4 vues

Transféré par Manasses junior

Pressure Vessel

- Full Text 01
- Fem notepad
- A15. Method for Multi-objective Optimized Designs of Surface Mount Permanent Magnet Motors with Concentrated or Distributed Stator.pdf
- 02_2011_077_Croccollo_04
- Fem Application Biomechanics
- INTRODUCTION TO FINITE ELEMENT METHOD.doc
- FEA_DEMO
- 10.1.1.832.5384
- FEA Summary
- 10.1.1.163.1989
- b17 Handout 4
- ijnaoe-2013-0226
- General Procedure of FEA 7
- Me Less Sssssssss Sss
- Will Poorly-Shaped Elements Really Affect My Solution_ _ CAE Associates
- 639-1862-1-PB
- Ansys Lab Manual
- IFEM.apporigins of FEM
- 4I16-IJAET0916854_v6_iss4_1464to1473.pdf
- Matlab-toolbook

Vous êtes sur la page 1sur 116

by

ARTIK PATEL

Presented to the

Faculty of the Graduate School of

The University of Texas at Arlington

in Partial Fulfillment of the Requirements

for the Degree of

November 2016

Copyright © by Artik Patel 2016

ii

ACKNOWLEDGEMENTS

First of all, I would like to express my deep gratitude and admiration to my advisor and mentor

Dr.Bo.P.Wang, who is a constant source of guidance and inspiration to me. I am very grateful to

you for the encouragement and support which you have provided throughout my time as your

student. I have been extremely lucky to have a supervisor who cared so much about my work,

and who responded to my questions and queries so promptly.

I want to extend my sincere appreciation and thanks to my committee members, Dr. Kent

Lawrence and Dr. Wen S. Chan, who have been extremely helpful during the whole process of

thesis review and have helped enrich my thesis with their valuable suggestions. I would like to

mention a special thanks to Professor Weiya Jin and Professor Mingjne Zhou from the Design

Institute of Chemical Machinery, Zhejiang University for providing me the pressure vessel model.

Their insight and expertise has greatly assisted this research. I would also like to thank my friends

for the encouragement they made in support of my research.

Most importantly I would like to dedicate this work to my parents who have been there for me

always and are responsible for what I am today and without whom, the joy of completing the

educational phase of my life would be incomplete. This accomplishment would not have been

possible without them. Finally, I would like to express my gratitude to one and all, who directly or

indirectly, have lent their hand in the support of my thesis work. Thank you.

iii

Design Optimization of Pressure Vessel with Particular Design

Considerations

By

Artik Patel

Mechanical Engineering

M.S – The University of Texas at Arlington

ABSTRACT

The pressure vessels in industries are generally designed with a high safety factor because the

rupture of a pressure vessel can be extremely dangerous. A vessel that is poorly designed or

ineffectively designed to handle high pressure pose a very significant threat to life and property.

Because of this, the design and verification of pressure vessels is governed by design codes

specified by the ASME (American Society of Mechanical Engineers) Boiler and Pressure Vessel

Code. The objective of this thesis work is to minimize the total weight of a real-world pressure

vessel structure subjected to stress constraints specified by the ASME section VIII division-2

code. Optimization is the process of finding the best feasible solution amongst the conventional

designs which accepts almost all designs which merely satisfies the problem requirements. The

main purpose of performing design optimization in pressure vessels is to reduce cost, by reducing

the weight with sufficient strength to avoid any modes of failure in the design. This work discusses

size optimization of axisymmetric pressure vessel considering an integrated approach in which

the optimization procedure is implemented by interfacing the commercial finite element analysis

software ANSYS with MATLAB optimization algorithm. A half model is used in conjunction with a

single-objective function that aims to minimize the total weight of the pressure vessel equipment.

Design parameters such as shell thickness and flange thickness are optimized while limiting the

maximum linearized membrane and membrane plus bending stresses below the ASME code

limits.

iv

Table of Contents

ACKNOWLEDGEMENTS ........................................................................................................................ iii

ABSTRACT ................................................................................................................................................ iv

LIST OF FIGURES................................................................................................................................... vii

LIST OF TABLES ....................................................................................................................................... x

CHAPTER 1: INTRODUCTION ............................................................................................................... 1

1.1 Introduction to Optimization ........................................................................................................... 1

1.2 MATLAB Optimization Tool ............................................................................................................ 1

1.3 Finite Element Analysis using ANSYS ......................................................................................... 1

1.4 Objective and Approach ................................................................................................................. 2

CHAPTER 2: LITERATURE SURVEY.................................................................................................... 3

2.1 History ............................................................................................................................................... 3

2.2 Literature Review ............................................................................................................................. 3

CHAPTER 3: Finite Element Analysis of the pressure vessel ............................................................ 5

3.1 Modelling of cylindrical pressure vessel ....................................................................................... 5

3.2 Material Properties and Element Type ......................................................................................... 8

3.2.1 SOLID95 Assumptions and Restrictions: ........................................................................... 10

3.3 Boundary Conditions ..................................................................................................................... 11

3.4 Loadings.......................................................................................................................................... 14

3.4.1 Gravity along X-axis ............................................................................................................... 14

3.4.2 Internal Pressure .................................................................................................................... 15

3.4.3 The average bolt pressure on the lower bolt surface of the two lateral flanges ........... 16

3.4.4 The average bolt pressure on the upper bolt surface of the two lateral flanges .......... 17

3.4.5 The average bolt pressure on the lower bolt surface of the two inner flanges ............. 18

3.4.6 The average bolt pressure on the upper bolt surface of the two inner flanges ............ 19

3.4.7 The pull on the cross section of the two end nozzles ....................................................... 20

3.5 Meshing ........................................................................................................................................... 21

3.5.1 Tetrahedral Mesh VS Hexahedral Mesh............................................................................. 21

3.5.2 Comparison between free mesh and controlled mesh ..................................................... 24

3.5.3 Mesh Convergence and Stress Singularity ........................................................................ 25

CHAPTER 4: Stress Analysis and Verification .................................................................................... 30

4.1 Introduction to ASME Boiler and Pressure Vessel – Verification Code ................................ 30

4.2 Stress Linearization ....................................................................................................................... 30

v

4.3 Stress Classification ...................................................................................................................... 32

4.4 Design Limits and Verifications ................................................................................................... 33

4.5 ANSYS Stress Linearization Results .......................................................................................... 35

4.5.1 Path-1 ....................................................................................................................................... 37

4.5.2 Path-2 ....................................................................................................................................... 39

4.5.3 Path-3 ....................................................................................................................................... 41

4.5.4 Path-4 ....................................................................................................................................... 43

4.5.5 Path-5 ....................................................................................................................................... 45

4.5.6 Path-6 ....................................................................................................................................... 47

4.5.7 Path-7 ....................................................................................................................................... 49

4.5.8 Path-8 ....................................................................................................................................... 51

4.5.9 Path-9 ....................................................................................................................................... 53

4.5.10 Path-10 .................................................................................................................................. 55

4.5.11 Path-11 .................................................................................................................................. 57

4.5.12 Path-12 .................................................................................................................................. 59

4.5.13 Path-13 .................................................................................................................................. 61

4.5.14 Path-14 .................................................................................................................................. 63

4.5.15 Paths-15,16,17,18,19 and 20 ............................................................................................. 65

4.6 Summary of Stress Analysis Results.......................................................................................... 72

CHAPTER 5: MATLAB Optimization ..................................................................................................... 73

5.1 Optimization Problem Formulation in MATLAB ........................................................................ 74

5.2 Design Space Exploration ............................................................................................................ 76

5.3 MATLAB fmincon function ............................................................................................................ 81

CHAPTER 6: Integration of ANSYS and MATLAB ............................................................................. 84

CHAPTER 7: Results and Conclusion .................................................................................................. 86

7.1 Results ............................................................................................................................................ 86

7.2 Conclusions .................................................................................................................................. 100

7.3 Future Work .................................................................................................................................. 101

REFERENCES ....................................................................................................................................... 102

APPENDIX .............................................................................................................................................. 104

BIBLIOGRAPHY ..................................................................................................................................... 105

vi

LIST OF FIGURES

Figure 2: Pressure Vessel full-model ............................................................................................................. 6

Figure 3:Main Components of the Pressure Vessel Equipment ................................................................... 7

Figure 4: Cross-sectional area of the vessel .................................................................................................. 8

Figure 5: Rotating cross-sectional area to generate the vessel volume ....................................................... 8

Figure 6: SOLID95 3-D 20-Node Structural Solid......................................................................................... 10

Figure 7: Symmetry constraint on the z-plane............................................................................................ 11

Figure 8: Fixed support on the position of 1st bearing ............................................................................... 12

Figure 9: Constraint along X-axis on the position of 2nd bearing............................................................... 13

Figure 10: Acceleration due to gravity ........................................................................................................ 14

Figure 11: Internal Pressure of 0.2 MPa ..................................................................................................... 15

Figure 12: Bolt pressure on the lower bolt surface of the two lateral flanges ........................................... 16

Figure 13: Bolt pressure on the upper bolt surface of the two lateral flanges .......................................... 17

Figure 14: Bolt pressure on the lower bolt surface of the two inner flanges ............................................. 18

Figure 15: Bolt pressure on the upper bolt surface of the two inner flanges ............................................ 19

Figure 16: The pull on the cross section of the two end nozzles ................................................................ 20

Figure 17: Tetra-meshed Model of the Pressure Vessel ............................................................................. 22

Figure 18: Unaveraged Von-Mises Stress Contour for Tetra-meshed Model............................................. 23

Figure 19: Brick-meshed Model of the Pressure Vessel ............................................................................. 23

Figure 20: Unaveraged Von-Mises Stress Contour for Hexa-meshed Model ............................................. 24

Figure 21: Free mesh with Brick elements.................................................................................................. 25

Figure 22: Controlled mesh with Brick elements ........................................................................................ 25

Figure 23: Stress Intensity for Free Hexahedral mesh ................................................................................ 25

Figure 24: Stress Intensity for Controlled Hexahedral mesh ...................................................................... 25

Figure 25: Mesh Convergence Plot ............................................................................................................. 27

Figure 26: Maximum Local Unaveraged Von-Mises stress ......................................................................... 27

Figure 27: Increase in Maximum Local Unaveraged Von-Mises stress for mesh refinement-1 ................. 28

Figure 28: Increase in Maximum Local Unaveraged Von-Mises stress for mesh refinement-2 ................. 28

Figure 29: Increase in Maximum Local Unaveraged Von-Mises stress for mesh refinement-3 ................. 29

Figure 30: Increase in Maximum Local Unaveraged Von-Mises stress for mesh refinement-4 ................. 29

vii

Figure 31: Path-1 plot on geometry ............................................................................................................ 37

Figure 32: Stress variation through the thickness along Path-1 ................................................................. 38

Figure 33: Path-2 plot on geometry ............................................................................................................ 39

Figure 34: Stress variation through the thickness along Path-2 ................................................................. 40

Figure 35: Path-3 plot on geometry ............................................................................................................ 41

Figure 36: Stress variation through the thickness along Path-3 ................................................................. 42

Figure 37: Path-4 plot on geometry ............................................................................................................ 43

Figure 38: Stress variation through the thickness along Path-4 ................................................................. 44

Figure 39: Path-5 plot on geometry ............................................................................................................ 45

Figure 40: Stress variation through the thickness along Path-5 ................................................................. 46

Figure 41: Path-6 plot on geometry ............................................................................................................ 47

Figure 42: Stress variation through the thickness along Path-6 ................................................................. 48

Figure 43: Path-7 plot on geometry ............................................................................................................ 49

Figure 44: Stress variation through the thickness along Path-7 ................................................................. 50

Figure 45: Path-8 plot on geometry ............................................................................................................ 51

Figure 46: Stress variation through the thickness along Path-8 ................................................................. 52

Figure 47: Path-9 plot on geometry ............................................................................................................ 53

Figure 48: Stress variation through the thickness along Path-9 ................................................................. 54

Figure 49: Path-10 plot on geometry .......................................................................................................... 55

Figure 50: Stress variation through the thickness along Path-10 ............................................................... 56

Figure 51: Path-11 plot on geometry .......................................................................................................... 57

Figure 52: Stress variation through the thickness along Path-11 ............................................................... 58

Figure 53: Path-12 plot on geometry .......................................................................................................... 59

Figure 54: Stress variation through the thickness along Path-12 ............................................................... 60

Figure 55: Path-13 plot on geometry .......................................................................................................... 61

Figure 56: Stress variation through the thickness along Path-13 ............................................................... 62

Figure 57: Path-14 plot on geometry .......................................................................................................... 63

Figure 58: Stress variation through the thickness along Path-14 ............................................................... 64

Figure 59: Additional paths created near the stress singularity region ...................................................... 65

Figure 60: Stress variation through the thickness along Path-15 ............................................................... 66

Figure 61: Stress variation through the thickness along Path-16 ............................................................... 67

Figure 62: Stress variation through the thickness along Path-17 ............................................................... 68

viii

Figure 63: Stress variation through the thickness along Path-18 ............................................................... 69

Figure 64: Stress variation through the thickness along Path-19 ............................................................... 70

Figure 65: Stress variation through the thickness along Path-20 ............................................................... 71

Figure 66: Design Variables ......................................................................................................................... 75

Figure 67: Response of the Pressure Vessel Model .................................................................................... 78

Figure 68: Design Space .............................................................................................................................. 79

Figure 69: Membrane Stress Contours ....................................................................................................... 80

Figure 70: Membrane Plus Bending Stress Contours ................................................................................. 80

Figure 71: Objective function(Volume) vs shell thickness .......................................................................... 81

Figure 72: Objective function(Volume) vs flange thickness........................................................................ 81

Figure 73: Integration Flow Chart ............................................................................................................... 86

Figure 74: fmincon run-1 iterations ............................................................................................................ 88

Figure 75: Plot functions for fmincon run-1................................................................................................ 88

Figure 76: fmincon run-2 iterations ............................................................................................................ 89

Figure 77: Plot Functions for fmincon run-2 ............................................................................................... 90

Figure 78: fmincon run-3 iterations ............................................................................................................ 91

Figure 79: Plot Functions for fmincon run-3 ............................................................................................... 92

Figure 80: fmincon run-4 iterations ............................................................................................................ 93

Figure 81: Plot functions for fmincon run-4................................................................................................ 93

Figure 82: fmincon run-5 iterations ............................................................................................................ 94

Figure 83: fmincon run-6 iterations ............................................................................................................ 95

Figure 84: Plot functions for fmincon run-6................................................................................................ 96

Figure 85: fmincon run-7 iterations ............................................................................................................ 97

Figure 86: Plot Functions for fmincon run-7 ............................................................................................... 98

Figure 87: fmincon run-8 iterations ............................................................................................................ 99

Figure 88: Plot Functions for fmincon run-8 ............................................................................................... 99

ix

LIST OF TABLES

Table 1: Chemical composition % of steel grade S30408 --------------------------------------------------------------- 9

Table 2: Design Stress Intensity of Material at different Temperatures -------------------------------------------- 9

Table 3: Material of the main pressure-bearing components of the pressure vessel --------------------------- 9

Table 4: Mechanical Properties of the Material -------------------------------------------------------------------------- 9

Table 5: Mesh Convergence Results ---------------------------------------------------------------------------------------- 26

Table 6: Stress Limits as per the ASME code------------------------------------------------------------------------------ 34

Table 7: Design Conditions for Pressure Vessel Equipment----------------------------------------------------------- 35

Table 8: Path-1 Evaluation ---------------------------------------------------------------------------------------------------- 37

Table 9: ANSYS Path-1 Linearized Results --------------------------------------------------------------------------------- 38

Table 10: Path-2 Evaluation --------------------------------------------------------------------------------------------------- 39

Table 11: ANSYS Path-2 Linearized Results ------------------------------------------------------------------------------- 40

Table 12: Path-3 Evaluation --------------------------------------------------------------------------------------------------- 41

Table 13: ANSYS Path-3 Linearized Results ------------------------------------------------------------------------------- 42

Table 14: Path-4 Evaluation --------------------------------------------------------------------------------------------------- 43

Table 15: ANSYS Path-4 Linearized Results ------------------------------------------------------------------------------- 44

Table 16: Path-5 Evaluation --------------------------------------------------------------------------------------------------- 45

Table 17: ANSYS Path-5 Linearized Results ------------------------------------------------------------------------------- 46

Table 18: Path-6 Evaluation --------------------------------------------------------------------------------------------------- 47

Table 19: ANSYS Path-6 Linearized Results ------------------------------------------------------------------------------- 48

Table 20: Path-7 Evaluation --------------------------------------------------------------------------------------------------- 49

Table 21: ANSYS Path-7 Linearized Results ------------------------------------------------------------------------------- 50

Table 22: Path-8 Evaluation --------------------------------------------------------------------------------------------------- 51

Table 23: ANSYS Path-8 Linearized Results ------------------------------------------------------------------------------- 52

Table 24: Path-9 Evaluation --------------------------------------------------------------------------------------------------- 53

Table 25: ANSYS Path-9 Linearized Results ------------------------------------------------------------------------------- 54

Table 26: Path-10 Evaluation ------------------------------------------------------------------------------------------------- 55

Table 27: ANSYS Path-10 Linearized Results ------------------------------------------------------------------------------ 56

Table 28: Path-11 Evaluation ------------------------------------------------------------------------------------------------- 57

Table 29: ANSYS Path-11 Linearized Results: ----------------------------------------------------------------------------- 58

Table 30: Path-12 Evaluation ------------------------------------------------------------------------------------------------- 59

x

Table 31: ANSYS Path-12 Linearized Results ------------------------------------------------------------------------------ 60

Table 32: Path-13 Evaluation ------------------------------------------------------------------------------------------------- 61

Table 33: ANSYS Path-13 Linearized Results ------------------------------------------------------------------------------ 62

Table 34: Path-14 Evaluation ------------------------------------------------------------------------------------------------- 63

Table 35: ANSYS Path-14 Linearized Results ------------------------------------------------------------------------------ 64

Table 36: Path-15 Evaluation ------------------------------------------------------------------------------------------------- 66

Table 37: Path-16 Evaluation ------------------------------------------------------------------------------------------------- 67

Table 38: Path-17 Evaluation ------------------------------------------------------------------------------------------------- 68

Table 39: Path-18 Evaluation ------------------------------------------------------------------------------------------------- 69

Table 40: Path-19 Evaluation ------------------------------------------------------------------------------------------------- 70

Table 41: Path-20 Evaluation ------------------------------------------------------------------------------------------------- 71

Table 42: Stress Verification -------------------------------------------------------------------------------------------------- 72

Table 43: fmincon Optimization Options ---------------------------------------------------------------------------------- 82

Table 44: fmincon Optimization Convergence for different step tolerance limits ---------------------------- 100

Table 45: Optimization Results Summary ------------------------------------------------------------------------------- 100

xi

CHAPTER 1: INTRODUCTION

Optimization in general can be defined as the act of finding the best feasible solution with the

most cost effective or highest achievable performance under the given constraints, by maximizing

desired factors and minimizing undesired ones. The increased demand to cut down the production

and manufacturing costs in the industries have encouraged engineers to use more robust decision

making technique such as optimization. In any optimization problem, we seek values of the design

parameters that minimize or maximize the objective while satisfying constraints.

Over a past few decades, the optimization techniques have found its applications in a wide variety

of industries such as automotive, aerospace, chemical, electrical and manufacturing industries.

Because of the advancement in computer technology, the complexity of a problem being solved

by optimization methods is no longer a concerned issue. Although this process can sometimes

be very time consuming, depending on the size and nature of problem in hand. With the advent

of computers, engineers can exploit and implement this procedure in practice. To fully realize the

power of computational Design Optimization, it is important to implement optimization methods

through pertinent computer-based mathematical tools.

To conduct optimization for the pressure vessel model, we use MATLAB software. MATLAB is a

computational modeling and coding tool that is powerful, easy to use, and one that is widely

applied in engineering and other fields. MATLAB is used worldwide to optimize both simple and

complex systems or designs with effectiveness and efficiency. It provides a variety of inbuilt

optimization functions like fmincon, fminsearch, genetic algorithm, etc which uses search

algorithms that are executed iteratively by comparing various solutions till an optimum or a

satisfactory solution is found. It is entirely dependent on the analyst to choose an appropriate

optimization function which is computationally efficient and accurate for the design problem in

hand. In this thesis work, we will restrict the analysis to fmincon function only. FMINCON is a

MatLab inbuilt nonlinear solver for optimization. It has proved to be a suitable tool to solve many

optimization problems in the mechanical engineering field.

A completely accurate representation of the physical model may lead to an extremely complex

mathematical model that may be hard to solve with the available hardware and software

resources. For this reason, we use a finite element model which is a mathematical representation

of a real-life component or system that is being analyzed. FEA consists of three main steps: Pre-

processing, solution and post-processing. Often in the engineering world, the structural analysis

1

is carried out using the Finite Elements Method. In this method, the physical model is discretized

into several small and simple parts called ‘finite elements’. The simple equations that model these

finite elements are then assembled into a larger system of equations that models the entire

problem. Finite elements are employed to determine the deformation and stresses in a structure

subjected to loads and boundary conditions. Mathematically it may be considered as a numerical

tool to analyze problems governed by partial differential equations that describe the behavior of

the system being studied.

There are a lot of commercial finite element analysis software available in the market such as

ANSYS, ABAQUS, ALTAIR HYPERWORK, NASTRAN, etc. In the present work, ‘ANSYS v17.0’

was used for finite element analysis of the pressure vessel model. ANSYS is one of the most

powerful engineering analysis software. It is widely used in the engineering industries to perform

finite element analysis, structural analysis, computational fluid dynamics, and heat transfer. This

computer simulation product provides finite elements to model behavior, and supports material

models and equation solvers for a wide range of mechanical design problems.

The finite element model of the pressure vessel was created using three dimensional solid

elements. An APDL (ANSYS Parametric Design Language) script file was written to automate the

pre-processing, solution and post-processing phase in ANSYS. The obtained stress intensity

results were linearized along 20 different stress classification lines(SCL/paths) to extract the

membrane and membrane plus bending stress generated in the different components of the

pressure vessel like nozzles, flanges cylindrical vessel body and the vessel head. These stresses

are then classified as primary or secondary, depending on the influence from stress singularity

region. Finally, both primary and secondary membrane and membrane plus bending stresses are

compared against their respective ASME code limits.

The main objective of this thesis work is to minimize the total weight of a Pressure vessel

subjected to stress constraints specified by the ASME ‘Design by Analysis of Boiler and Pressure

Vessel’ code limits. This is achieved by using FEA results obtained from ANSYS in conjunction

with MATLAB fmincon optimization solver for the sole purpose of minimizing the objective. Solving

complex design problems by integrating robust optimization tools like MATLAB with powerful FEA

softwares such as ANSYS has opened a new door in the field of design optimization. Optimization

methods, combined with more detailed and accurate simulation methods can improve the

experimental process of conceptual and detailed design of engineering systems.

2

Finite Element Analysis of the pressure vessel model is executed through MATLAB by running

ANSYS script file in Batch mode. The post-processing results such as total volume of the

equipment, deformation, von-mises stress and linearized membrane and membrane plus bending

stresses are stored in a text file. Objective function and constraints are defined in MATLAB by

extracting appropriate data from this result file generated by ANSYS. The volume data gives the

value of the objective function, whereas the linearized stress results which are verified against

their ASME allowable values, forms the constraint. Finally, the MATLAB optimization algorithm

evaluates the objective function iteratively by comparing various solutions till an optimum design

is found. The proposed methodology is completely automated and does not require any kind of

user intervention until the optimal solution is found.

2.1 History

In the industrial sector, the pressure vessels are used as storage tanks, diving cylinder,

recompression chamber, distillation towers, autoclaves and many other vessels in mining or oil

refineries and petrochemical plants, nuclear reactor vessel, habitat of a space ship, habitat of a

submarine, pneumatic reservoir, hydraulic reservoir under pressure, rail vehicle airbrake

reservoir, road vehicle airbrake reservoir and storage vessels for liquefied gases. The design

analysis of pressure vessels is an important and practical topic which has been investigated for

decades. Even though optimization techniques have been extensively applied to design

structures in general, very few pieces of work can be found which are directly related to

optimization of pressure vessels by interfacing different software packages like MATLAB and

ANSYS. These few references include the design optimization of homogeneous as well as

composite pressure vessels with different optimization methods.

In the paper 'Integration of MATLAB and ANSYS for Advanced Analysis of Vehicle Structures',

A.Gauchia and B.L.Boada has explained the optimization of a complex bus structure in weight

and stiffness by means of coupling MATLAB and ANSYS. For the optimization loop analyzed in

this study the genetic algorithm toolbox has been employed, having shown to be a very useful

tool. A reduction of 4% of the weight was achieved while improving the torsion stiffness in 0,23%.

Prior to this optimization, a sensitivity analysis was carried out in order to apply the optimization

loop on certain beams more sensitive to variations in weight and torsion stiffness. [1]

3

In the work by Levi B. de Albuquerque and Miguel Mattar Neto, design criteria were developed to

preclude the various pressure vessel failure modes through the so-called "Design by Analysis"

method. In the "Design by Analysis" approach, also used in Section VIII, Division 2 of the Code,

the design limits were established in correspondence to each failure mode. A typical Pressurized

Water Reactor (PWR) nozzle to pressure vessel connection subjected to internal pressure and

concentrated loads was modeled with 3D solid finite elements in linear elastic and limit load

analyses. Using some stress categorization approaches, the results from linear elastic and limit

load analyses were compared to each other and also with results obtained by formulae for simple

shell geometries. Based on the result comparison, some conclusions and recommendations on

the type of finite element analysis (linear elastic or limit load) and on the stress categorization

were addressed for the studied cases. [2]

The research conducted by Carlos A. de J. Miranda, Altair A. Faloppa, Miguel Mattar Neto and

Gerson Fainer shows a discussion on how to perform the stress verifications based on a generic

geometry found in many plants, from petrochemical to nuclear. In this study, the author discusses

the nuclear piping analysis with a non-standard item when the item should be modeled as a 3D

solid with its verification done per the Sub-section NB 3300 of the ASME Code. Only the primary

stresses due to the internal pressure were considered since the scope of the work was to

emphasize some of the issues that arise from the stress classification and linearization in

discontinuities, which are common in the nuclear area. Along with the modeling, analysis and

verification a discussion on how to perform the Code verifications was presented, pointing some

differences between the present(simplified) analysis, just one load – pressure, and an actual one,

with several applied loads. [3]

axisymmetric pressure vessels considering an integrated approach in which the entire pressure

vessel model is used in conjunction with a multi-objective function that aims to minimize the von-

Mises mechanical stress from nozzle to head. Representative examples are examined and

solutions obtained for the entire vessel considering temperature and pressure loading [4].

The paper submitted on 'Design & Weight Optimization of Pressure Vessel Due to Thickness

Using Finite Element Analysis' by Vishal V.Saidpatil and Arun S.Thakare explains the detailed

design & analysis of Pressure vessel used in boiler for optimum thickness, temperature

distribution and dynamic behavior using ANSYS. Their work involves design of a cylindrical

4

pressure vessel to sustain 5 bar pressure and determine the wall thickness required for the vessel

to limit the maximum shear stress [5].

Sulaiman Hassan and Kavi Kumar considered a metaheuristic approach to optimize the pressure

vessel design. The work parameters such as thickness of the shell, and dish end, length and

radius of the pressure vessel were optimized by making use of Ant colony optimization (ACO)

Algorithm. They found that the results obtained from ACO are better as its search is for global

optimum as against the local optimum in traditional search methods [6].

K. Sahitya Raju and Dr. S. Srinivas Rao conducted Design optimization of a composite cylindrical

pressure vessel using FEA. In this work, design analysis of fiber reinforced multi layered

composite shell, with optimum fiber orientations; minimum mass under strength constraints for a

cylinder under axial loading for static and buckling analysis on the pressure vessel has been

studied. It involves the comparison of conventional steel and Composite material cylindrical

pressure vessel under static loading conditions. [7].

A very few research is found that directly relates to the optimization of pressure vessel by

interfacing FEA software with an optimization tool. Many other researches found including

analytical, experimental and numerical investigations have been devoted to the design

optimization of head and nozzle connections in pressure vessels subjected to different external

loadings.

An ANSYS command file was written based on the design data of a real-world pressure vessel

equipment provided by the ‘Design Institute of Chemical Machinery’, Zhejiang University. This

script file is executed in ANSYS APDL v17.0 to generate the geometrical entities such as

keypoints, lines, areas and volumes. The ANSYS model as shown in figure-1 below, represents

an axisymmetric cylindrical pressure vessel used in chemical industries, that is approximately 6

meter-long and 2 meter-wide, with elliptical heads and four rectangular nozzles supported by the

flange. Due to symmetry of the vessel along the z-plane, we consider only half model for our

analysis. This saves a lot of computational time during the FEA as well as during the optimization

process. A full model displayed in figure-2 can be used for FEA, in case, when large amount of

memory(RAM) along with sufficient hardware and software resources are available, or when the

computational time is not an issue.

5

Figure 1: Pressure Vessel half-model Figure 2: Pressure Vessel full-model

As shown in figure-4, the geometry of the pressure vessel is defined based on the parameters

such as shell radius, shell height, shell thickness, head height etc. A quarter model is created by

rotating the cross-sectional area of the pressure vessel by 180 degrees about the y-axis.

Symmetricity of the structure was fully exploited by mirroring the volumes created in the quarter

model about the x-z plane to generate half model (figure-1). Nozzle and flange supports were

created with the help of ANSYS pre-processing functions like extrude and volume delete. The

sharp corners on nozzle edges and on flange edges have been filleted to reduce the stress

concentration around these corners. Note that, the lateral nozzle openings are longer than the

medial nozzle openings. The complete structure was subdivided into volume blocks to satisfy the

conditions of hexa-meshing to generate brick elements throughout the model. One more reason

to divide the volume blocks is to apply constraints in some specified location of the geometry.

6

The main Pressure bearing components of this pressure vessel equipment are shown in figure-3

below.

7

Figure 4: Cross-sectional area of the vessel Figure 5: Rotating cross-sectional area to generate

the vessel volume

Steel alloy ‘S30408’ standard ‘GB24511’ is the material used to create the pressure vessel

equipment. In general, this material is extensively used in Chinese steel industries to produce

products such as steel oils, sheets, plates, round bars, steel wires, pipes, forgings etc. GB

standards are the Chinese national standards issued by the Standardization Administration of

China (SAC), the Chinese National Committee of the ISO and the IEC. This Chinese standard

specifies classification and designation, dimensions, shapes and tolerances, technical

requirements, test methods, inspection rules, package, marks and product quality certificates of

Stainless steel plate, sheet and strip for pressure equipments. This standard applies to width of

not less than 600mm of pressure equipment with hot-rolled, cold-rolled stainless steel sheet and

8

strip. The chemical composition of this steel alloy is displayed below in table 1. The design stress

intensity of the materials at different temperatures is shown in Table 2. The material of the main

pressure-bearing components of this equipment is shown in Table 3 and the mechanical

properties are presented in Table 4.

Normal

Design Stress(MPa) at different

Thickness Temperature

Steel Type Standard Temperatures (℃)

(mm) Strength (MPa)

Rm R el <20 100 150 200 250 300

S30408 GB 24511 1.5~80

Plate 520 205 137 114 103 96 90 85

Steel

S3408 GB 13296 ≤13 520 205 137 137 137 130 122 114

Pipe

1 Case S30408 GB 24511

2 Forming Head S30408 GB 24511

3 Support Q235A/S3408 GB/T3274{GB150.2} / GB 24511

4 Flange S30408 GB 24511

PROPERTY VALUE

Young’s Modulus 200 GPa

Poisson’s Ratio 0.3

Density 7.9e-6 𝐾𝑔/𝑚𝑚3

Yield Strength 345 MPa

9

In ANSYS software, the user can select from over 100 different element types to construct their

model. Solid95 20-nodes hexahedral/brick element was used for finite element analysis of the

pressure vessel model. As shown in figure-6 below, Solid95 is a 3-dimensional solid element. It

can tolerate irregular shapes without as much loss of accuracy. Solid95 elements have compatible

displacement shapes and are well suited to model curved boundaries. The hexahedral/Brick

element is defined by 20 nodes (including mid-nodes) whereas the tetrahedral element consists

of 10 nodes (including mid-node). Each node has three degrees of freedom: translations in the

nodal x, y, and z directions. Therefore, each brick element consists of 60 DOF. The element may

have any spatial orientation. The element has plasticity, creep, stress stiffening, large deflection,

and large strain capabilities.

The element must not have a zero volume.

The element may not be twisted such that the element has two separate volumes.

This occurs most frequently when the element is not numbered properly.

may have the planes IJKL and MNOP interchanged.

10

An edge with a removed midside node implies that the displacement varies linearly,

rather than parabolically, along that edge.

Degeneration to the form of pyramid should be used with caution. The element sizes,

when degenerated, should be small to minimize the stress gradients.

Three different boundary conditions are imposed on the design model:

1) Symmetry boundary condition is applied to the structural symmetry plane (all area at Z=0).

2) Fixed support in X-axis and Y-axis on the position of one bearing. (fig-8)

11

Figure 8: Fixed support on the position of 1st bearing

12

Figure 9: Constraint along X-axis on the position of 2nd bearing

13

3.4 Loadings

Seven different loadings conditions are applied on the pressure vessel model:

2) Internal pressure of 0.2 MPa. (figure-11)

3) The average bolt pressure on the lower bolt surface (surface near the cylinder) of the two

lateral flanges. (figure-12)

4) The average bolt pressure on the upper bolt surface (surface far away from the cylinder)

of the two lateral flanges. (figure-13)

5) The average bolt pressure on the lower bolt surface (surface near the cylinder) of the two

inner flanges. (figure-14)

6) The average bolt pressure on the upper bolt surface (surface far away from the cylinder)

of the two inner flanges. (figure-15)

7) The pull on the cross section of the two end nozzles. (figure-16)

Gravity in negative X-axis, -9.8 m/s2 (So it is acceleration in forward of X-axis, 9.8)

14

3.4.2 Internal Pressure

The Pressure Vessel holds an internal pressure of 0.2 MPa.

15

3.4.3 The average bolt pressure on the lower bolt surface of the two lateral

flanges

The average bolt pressure on the lower bolt surface (near the cylinder) of the two lateral flanges.

On one flange, the total force is 193623 N, the area is 78100 mm2, the pressure is 2.479162149

MPa.

Figure 12: Bolt pressure on the lower bolt surface of the two lateral flanges

16

3.4.4 The average bolt pressure on the upper bolt surface of the two lateral

flanges

The average bolt pressure on the upper bolt surface (far away from the cylinder) of the two lateral

flanges. On one flange, the total force is 19397 N, the area is 78100 mm2, the pressure is

0.2483645393 MPa.

Figure 13: Bolt pressure on the upper bolt surface of the two lateral flanges

17

3.4.5 The average bolt pressure on the lower bolt surface of the two inner flanges

The average bolt pressure on the lower bolt surface (near the cylinder) of the two inner flanges.

On one flange, the total force is 143408 N, the area is 67740 mm2, the pressure is 2.117037053

MPa.

Figure 14: Bolt pressure on the lower bolt surface of the two inner flanges

18

3.4.6 The average bolt pressure on the upper bolt surface of the two inner flanges

The average bolt pressure on the upper bolt surface (far away from the cylinder) of the two inner

flanges. On one flange, the total force is 19400 N, the area is 67740 mm2, the pressure is

0.2863877948 MPa.

Figure 15: Bolt pressure on the upper bolt surface of the two inner flanges

19

3.4.7 The pull on the cross section of the two end nozzles

The pull on the cross section of the two end nozzles (Compensate for the gas pressure on the

cross section). On one nozzle, the total force is -40856 N, the area is 16336 mm2, the pressure

is -2.500961538 MPa.

Figure 16: The pull on the cross section of the two end nozzles

20

3.5 Meshing

Meshing is one of the most important aspect of finite element analysis. The accuracy of the FEA

results predominantly depends on the mesh size and element quality. The larger the density of

meshing, the greater is the accuracy of the geometry and greater is the difficulty in solving the

problems. Therefore, a preferred meshing approach is to employ fine meshes only in the area of

focus whereas larger meshes should be used in the region where we expect relatively low activity.

The pattern and relative positioning of the nodes also affect the solution, the computational

efficiency & time. This is why good meshing is very essential for a sound computer simulation to

give good results.

For the pressure vessel model, besides four volume block, the entire model was meshed with 20-

node Brick elements. Since these four volumes does not meet the hexa-meshing criteria, they

were meshed with 10-node tetrahedral elements. The ANSYS software automatically uses

pyramid elements as filler elements in between the mesh transition zones. The Hexahedral

meshed model of the pressure vessel used during optimization is shown below in figure-19.

Unlike tetrahedron meshing that can be performed on nearly any geometry, hex meshing (Brick

elements) requires a certain amount of topology cleaning and decomposition to achieve an all or

nearly all brick mesh. This type of meshing is generally preferred when less nodes and elements

are required but need to achieve high solution accuracy. A brick meshed model can save orders

of magnitudes of CPU time and require significantly less RAM and disk space over an all

tetrahedron mesh with often better accuracy. But the downside of Brick meshing or a hexahedral

mesh is that it is very difficult to generate for a complex geometry because it requires map

meshable sides to sweep through the volume.

The figure-17 displayed below shows a tetra mesh of the pressure vessel model. Ten node

tetrahedral elements were used to produce the tetra mesh model. For free meshing, a smart sizing

level-2 was set to obtain a very fine mesh with better element quality. Whereas, a hexa-mesh

model was produced by sweep meshing the volumes. The total vessel volume was split into

different blocks for more control and to meet topology requirement of map meshable sides. The

element sizes generated on swept volume were defined by assigning line divisions, taking into

account curvature of the line, its proximity to holes, element order and other features. Below

mentioned guideline was followed to achieve the volume sweep.

21

1. The source and target faces for all sweepable bodies are automatically detected by the

ANSYS software. If desired, the user can specify the source/target faces manually.

2. All source/target face topology needs to be same for all sources/targets.

3. All side faces need to be able to be mapped meshed.

It is evident from the figures below, that the tetra meshed model contain eight times more

elements compared to the number of elements in brick meshed model. The stress results

obtained from both the meshes are more or less similar but the time taken to solve the tetra-

meshed model is about five times more when compared to the computational time taken to solve

the brick meshed model.

22

Figure 18: Unaveraged Von-Mises Stress Contour for Tetra-meshed Model

23

Figure 20: Unaveraged Von-Mises Stress Contour for Hexa-meshed Model

The figure 21 to 24 below displays the comparison between a free and a controlled brick mesh of

the pressure vessel model. During free mesh, line divisions are not considered and the software

automatically determines the size and number of elements generated through the swept volumes.

Whereas, the controlled mesh is created by specifying number of line divisions on the sweepable

faces of the model geometry to produce different element sizes on different components of the

pressure vessel. Latter is preferred as it gives more control over the mesh size and element

quality. Note that, the free mesh generated by ANSYS, has less number of elements when

compared to elements in controlled mesh model. Although controlled mesh increases the number

of brick elements, the computational time taken to solve this model in a 16GB RAM system is

approximately same when compared to the time taken to solve free mesh model. We also achieve

a high solution accuracy due to the increased number of elements. Hence, controlled brick

meshed model is preferred for the finite element analysis as well as during the optimization run.

24

Figure 21: Free mesh with Brick elements Figure 22: Controlled mesh with Brick elements

Figure 23: Stress Intensity for Free Hexahedral mesh Figure 24: Stress Intensity for Controlled Hexahedral mesh

A mesh convergence study when performing Stress Analysis is necessary to instill confidence in

FEM results from the standpoint of mathematics. As we progressively refine the mesh, the size

of the elements reduces, which theoretically increases the solution accuracy and given enough

iterations it converges towards a specific result. If there is an analytical solution for the given

problem, the mesh refinement procedure will converge towards the exact solution. As mesh

elements decrease in size but increase in quantity, the computational requirements to solve a

given model increase. As mesh elements decrease in size, they reach a point of diminishing

returns on the level of accuracy compared to the computational overhead and time required to

compute the result. This means that a simulation requires much more time to compute the results,

but the result may change by an insignificant value. Hence, to overcome this problem, mesh

25

convergence study is performed to determine a mesh with minimum number of elements required

to maintain a satisfactorily balance between accuracy and computing resources.

However, as shown in table-5 and figure-25 below, the mesh convergence study when applied to

the 20-node brick meshed pressure vessel model resulted in non-convergence. The stress

solution does not converge with mesh refinement because of the stress singularity present at the

joint corner of cylindrical shell body and nozzle. A stress singularity is a point of the mesh where

the stress does not convergence towards a specific value. As we keep refining the mesh, the

stress at this point keeps on increasing. Theoretically, the stress at the singularity is infinite.

Typical situations where stress singularities occur are the appliance of a point load, sharp re-

entrant corners, corners of bodies in contact and point restraints. These singularities occurs often

in pressure vessel and boiler designs is practically unavoidable.

Since the pressure vessel model is analyzed as whole structure, the stress singularity at the

filleted corner is of importance. The mesh around this region is refined locally to capture the effects

of high stress concentration. Despite of removing sharp re-entrant edges by filleting the corners,

the stress concentration around these corners increases with increase in the elements. However,

displacement solution does converge to a value of 1.56 mm. Through path operations for critical

stress concentration lines at stress singularity region, we can predict probable value of true stress

at the elements near this singularity. While running the optimization loop, mesh density of the

pressure vessel model was kept fixed and the stresses evaluated near the singularity were

verified against the allowable local stress limits. The path operations and stress limits are

discussed in the next chapter. The figures 26-30 displayed below, shows the increase in

maximum local unaveraged von-mises stress with the increase in number of elements at the

corner of the nozzle and shell body joint.

No. of Elements

Mises Stress (MPa)

66686 325.76

95363 339.37

110604 381.94

143492 437.58

170770 449.09

26

Figure 25: Mesh Convergence Plot

27

Figure 27: Increase in Maximum Local Unaveraged Von-Mises stress for mesh refinement-1

Figure 28: Increase in Maximum Local Unaveraged Von-Mises stress for mesh refinement-2

28

Figure 29: Increase in Maximum Local Unaveraged Von-Mises stress for mesh refinement-3

Figure 30: Increase in Maximum Local Unaveraged Von-Mises stress for mesh refinement-4

29

CHAPTER 4: Stress Analysis and Verification

Relevant Codes of Practice, Industry Standard and/or Statement of Assessment Criteria:

ASME Boiler and Pressure Vessel Code, Section VIII, Rules for Construction of Pressure

Vessels, Division 2

Boilers and pressure vessels are used worldwide in various industries. They are naturally present

in the power engineering and gas engineering sectors. In order to ensure the safety and

operational efficiency of these vessels, necessary legal regulations have been developed by the

American Society of Mechanical Engineers (ASME). These regulations constitute the basis for

design and manufacture of these equipments.

One of the most important goal in a pressure vessel design is to assure safe and satisfactory

performance of the vessel. The ASME pressure vessel code is based on the observed safety of

vessels. The observations were turned into design rules and the vessels became safer through

their efforts. It is a real world working standard – its roots were born of failed vessels and dead

operators in an era long before concepts like stress concentrations were even known.

The ASME ‘Design by analysis’ code, particularly in its Section VIII division-2, has specific

requirements on how to assess the results from the stress analyses to make the necessary

verifications to avoid failure. They wrote the VIII-2 rules and developed the stress linearization

method as a guideline to check for the safe design. This division covers the mandatory

requirements, specific prohibitions and nonmandatory guidance for materials, design, fabrication,

inspection and testing, markings and reports, overpressure protection and certification of pressure

vessels having an internal or external pressure which exceeds 103 KPa. These requirements

apply to those equipments which are part of the pressure boundary for example valves, pumps,

pressure vessels, piping, etc. In this work, only the maximum allowable stress verifications were

made as per the ASME code. Stress linearization method was implemented to verify the design

of pressure vessel equipment under study. The details of the ASME codes are not included in this

report. The reader is encouraged to the read the ASME related reference material mentioned at

the end of this report.

Generally, the prototypes and models used in the analyses are developed with 2D plane or 3D

solid finite elements, and membrane and bending stresses cannot be evaluated directly from the

FEA results for these types of elements. Due to this fact, no direct comparison with the code limits

30

can be done and, besides that, the commercial finite element software’s like ALTAIR, ANSYS,

ABAQUAS, etc do not distinguish between primary and secondary stresses. Therefore, to

implement the required ASME Pressure Vessel and Boiler Code stress verifications for our finite

element model, we should perform stress linearization to extract the membrane and bending

stresses from the 3D solid model and, also, should classify these stress components as primary

and secondary for the purpose of stress verification against the ASME allowable limits.

decomposes a basically parabolic distribution into a uniform value (membrane stress), a linearly

varying value (bending stress), and possibly an extra component (peak stress). The stress

linearization is performed along a stress classification line. The stress classification line (SCL) are

created to linearize the stresses along a line, usually cutting through the thickness of the

component. A Stress Classification Line or SCL is a straight line defined by two nodes/points,

usually more or less perpendicular to both the inside and outside surfaces. Stress components

through the section/SCL are linearized by a line integral method and are separated into constant

membrane stresses, bending stresses varying linearly between end points, and peak stresses

(defined as the difference between the maximum and minimum principal stress). The stress

linearization tool takes the nodal data for the complex stress pattern found along this line and

breaks it down into membrane and bending stress components.

Stress Linearization is used to comply with design codes and requirements of the pressure vessel

industry. However, applicability of the utility is not limited to pressure vessels. You can use this

method to graph local stress tensors along a linear path and/or to determine the relative

contributions of bending and membrane stress for any type of structure. Stress linearization is not

required in the models with beam or shell elements because these elements naturally give the

stresses separated in membrane and bending components.

Note:

Stress linearization is available for brick, tetrahedral, plate, shell, and 2D elements, with or

without mid-side nodes.

Stress linearization is available for all linear and nonlinear analysis types that produce stress

results.

31

4.3 Stress Classification

The purpose of stress classification is to identify the Primary(P) and the Secondary(Q) stresses.

Primary stresses are defined as the stresses developed by an imposed loading that is necessary

to satisfy the laws of equilibrium in terms of the external and internal forces and moments.

Secondary stresses are the stresses that are developed by constraints due to geometrical

discontinuities and self-constraint. The classification of stresses into primary and secondary

categories separates the issues regarding overall strength, which is of primary importance and

therefore referred to the realm of primary stresses, from the issues of local behavior, which is of

secondary importance and therefore referred to the realm of secondary stresses.

It should be acknowledged that different kinds of stress have different degrees of significance and

thus should have different safety implications. For example, the objective of primary stress limits

is to prevent the loss of load-carrying capacity of the vessel, which is referred to as collapse

whereas type of failure that a secondary stress may cause is ratcheting or incremental collapse.

Hence, it is necessary to classify the stresses into different categories. The stress categories of

interest for the design analysis of our pressure vessel model are the primary stress, and its

subcategories of general and local primary membrane and bending stress, and the secondary

stresses. The peak stress is related to the assessment of fatigue failure of the material and will

not be used in our analysis.

For design purposes, the primary membrane stress is further divided into general primary

membrane stress and local primary membrane stress subcategories. The average value acting

on the whole section/line that is equivalent to the net force acting in the section due to the actual

stress distribution will be classified as Pm or PL depending on the distance of the section from the

discontinuity: Pm for those far sections and PL otherwise. This PL classification is justified because

there is a secondary ‘aspect’ in this stress near a discontinuity even if it comes from a mechanical

load.

The maximum value of the linear stress distribution which produces a net bending moment

equivalent to the moment produced by the actual stress distribution is called ‘bending stress’ Pb .

For mechanical loads, if the section is near a discontinuity this stress component is classified as

secondary, ‘Q’. The difference between the actual stress distribution and the sum of the average

and linear (membrane + bending) stress distributions give an equilibrated stress distribution.

32

PL – Localized Primary Membrane Stress

F – Peak Stress

Q – Secondary Stresses

These steps, the stress classification and the stress linearization, are not straightforward ones

and needs some ‘engineering’ judgment to choose the right section to evaluate the stresses in

discontinuities. This task, most of the time is not a simple one due to the nature of the involved

load and/or the complex geometry under analysis. In fact, there are several studies discussing on

how to perform these stress classification and linearization.

ASME Section VIII-2 provides a guide to what the maximum stresses are allowed for different

locations of the pressure vessel. The combination of this ASME code and the output from the

stress linearization and classification tool is used to produce pass fail judgments on the pressure

vessel model. This will form the basis of constraint function in the optimization process.

As the ASME limits are developed aiming to prevent some typical failure modes besides the

Primary and Secondary classification, the stresses should be linearized to obtain the generalized

(Pm ) or localized (PL ) membrane component, the bending (Pb ) and the Peak (F) stress. Because

different modes of failure are associated with primary membrane, primary bending and secondary

stress, different allowable values are defined for each category. These are not given as absolute

values in the pressure vessel codes, but as a proportion of the basic allowable stress intensity of

the material (Sm) at design/working temperature. For the pressure vessel model in hand, standard

steel S30408 was used, which has a basic allowable stress intensity value of 137 MPa at the

working temperature of 20°C - 150°C.

1) General Primary Membrane Stress Intensity (Pm)

2) Local Primary Membrane Stress Intensity (PL)

3) Primary Membrane Plus Primary Bending Stress (PL + Pb) – either General or Local

Membrane Stress

33

4) Primary Plus Secondary Stress Intensity (PL + Pb + Q)

5) Peak Stress Intensity (PL + Pb + Q + F)

According to the ASME code, the maximum allowable stress limits for the pressure vessel model

in consideration are shown below:

Table 6: Stress Limits as per the ASME code

Design Stress Intensity (Sm): Basic allowable stress intensity of the material at design/working

temperature. Typically, the lesser of 2/3 the Yield Stress (YS) or 1/3 of the Ultimate Tensile Stress

(UTS).

Pm ≤ Sm = 137 MPa

The above indicated limits along with the equipment operating conditions shown in table-7 are

the so-called Design Condition. When the Operational Conditions are verified the pressure and

temperature are lower, but the earthquake should be considered, the limits are slightly different

as well as the Sm value (which depends on the temperature).

34

Primary generalized membrane stresses are not allowed to exceed the basic allowable stress

Sm, otherwise there is the possibility of a catastrophic plastic collapse e.g. a burst under pressure.

For the primary localized membrane stress, a margin of safety is included by specifying an

allowable membrane stress of 1.5×Sm. The total primary (membrane plus bending) allowable

stress has an allowable limit of 1.5×Sm. Secondary stress can comfortably exceed the material

allowable limit but must be limited to ensure shakedown under cyclic load. Hence the range of

secondary stress is limited to 3×Sm. Local stresses around nozzles or transitions could be higher

than global stresses – sometimes 2x as high, depending on the location and cause.

In its post-processor module, the ANSYS program can linearize the stresses along a given section

defined by two nodes –the SCL(Path). It linearizes all six stress components (SX, SY, SZ, SXY,

SYZ, SXZ). Also, the Tresca (SINT) and von Mises (SEQV) equivalent stresses are reported by

ANSYS software. To perform the linearization, the program considers 47 internal points along the

SCL. With this procedure, the Membrane (average), the Bending (linear), the Membrane ±

Bending, the Peak stress, the total stress, are calculated.

35

The stress linearization was performed on the nominal design of the pressure vessel model with

shell thickness of 10 mm and flange thickness of 16 mm. 20 sections or SCL were chosen to

linearize and classify the stresses, to cover all critical parts and regions in the analyzed geometry

of the pressure vessel, aiming to verify them against the Code limits. These sections are shown

in figures below where they are named PATH-X where X is the stress classification line number.

Using the ANSYS post processor, the linearized stresses along the defined classification lines

are extracted. The Tresca equivalent stress or the averaged stress intensity is used for

linearization. This is given directly by the software so it is not required to do the calculations

manually. The software first linearizes the stresses at a component level and then calculates the

equivalent stress on the results. The stresses are then classified as necessary and the linearized

stresses are checked against the allowable stress limits. Two elements throughout the shell of

the pressure vessel has been used during the linearization process.

It may be noted that the membrane plus bending plot is not linear across the section thickness for

some of the evaluations, depending on the stress location. However, the graphs shown are for

the equivalent stress intensity which due to the nature of its calculation will result in the contours

shown. ANSYS lists both the component linearized stresses and the calculated Tresca’s and von

Mises’ equivalent stress. The results are grouped by type, namely; membrane, bending,

membrane plus bending, peak and total. The tables below, list the linearization results for all the

20 stress classification lines defined in ANSYS. The maximum membrane and membrane plus

bending stress along all these paths are listed in the path evaluation table. For each path, the

table also shows the assigned stress categories, allowable and calculated stresses.

36

4.5.1 Path-1

This path is created along the thickness of the cylindrical shell where the shell body of the

pressure vessel connects to the Top-Head of the vessel. Since this path lies away from the

discontinuity, the average membrane stress and membrane plus bending stress along this path

are classified as primary and hence checked against the 1.5×Sm limit.

Equivalent

Allowable ANSYS Stress

Load Stress Stress Control Evaluation

Stress 𝐒𝐦 linearization Type Classification

Condition Intensity Value Result

(MPa) of stress symbol

SINT (MPa)

SI (Pm ) N/A 1.0×Sm =137 N/A

Design MEMBRANE

SII (PL ) 59.85 1.5×Sm =205.5 PASS

Conditions 137

A MEMBRANE PLUS

SIII (PL + Pb ) 73.26 1.5×Sm =205.5 PASS

BENDING

37

Figure 32: Stress variation through the thickness along Path-1

38

4.5.2 Path-2

Path-2 is the path along cylindrical shell thickness at the corner where the shell body of the vessel

connects rectangular Nozzle-1. Due to singularity around the corner, the stresses linearized along

this path are influenced by the local stress concentration effect. As the result of this discontinuity,

the total membrane plus bending stress near this region also comprises of some secondary

stresses, which needs to be considered when comparing with the allowable limits. Thus, total

membrane plus bending stress in this region is the sum of primary stress ‘PL + Pb ’and secondary

stress ‘Q’. Due to the presence of this secondary stress component, the allowable limit for total

membrane plus bending stress along this path is set to three times the basic allowable limit ‘Sm’.

Equivalent

Allowable ANSYS Stress

Load Stress Stress Control Evaluation

Stress 𝐒𝐦 linearization Type Classification

Condition Intensity Value Result

(MPa) of stress symbol

SINT (MPa)

SI (Pm ) N/A 1.0×Sm =137 N/A

Design MEMBRANE

Conditions 137 SII (PL ) 105.1 1.5×Sm =205.5 PASS

A MEMBRANE PLUS

𝑆𝐼𝑉 (𝑃𝐿 + 𝑃𝑏 + 𝑄 ) 206.6 3×Sm =411 PASS

BENDING

39

Figure 34: Stress variation through the thickness along Path-2

40

4.5.3 Path-3

Path-3 is the path along Rectangular Nozzle thickness at the corner where the pressure vessel

shell body connects rectangular Nozzle N-1. Since this path lies very close to the singularity at

corner, the secondary stresses need to be considered for this path. The total membrane plus

bending stress in this region is the sum of primary stress ‘PL + Pb ’and secondary stress ‘Q’.

Therefore, the allowable limit for total membrane plus bending stress along this path is three times

the basic allowable limit ‘Sm’.

Equivalent

Allowable ANSYS Stress

Load Stress Stress Control Evaluation

Stress 𝐒𝐦 linearization Type Classification

Condition Intensity Value Result

(MPa) of stress symbol

SINT (MPa)

SI (Pm ) N/A 1.0×Sm =137 N/A

Design MEMBRANE

Conditions 137 SII (PL ) 157.1 1.5×Sm =205.5 PASS

A MEMBRANE PLUS

𝑆𝐼𝑉 (𝑃𝐿 + 𝑃𝑏 + 𝑄 ) 235.0 3×Sm =411 PASS

BENDING

41

Figure 36: Stress variation through the thickness along Path-3

42

4.5.4 Path-4

Path-4 is the path along the thickness of the rectangular Nozzle at the middle section where

cylindrical shell body of the vessel connects the Nozzle N-1. This path is far from the singularity.

The average membrane stress and membrane plus bending stress along this path are classified

as primary and therefore, are checked against the 1.5×Sm limit.

Equivalent

Allowable ANSYS Stress

Load Stress Stress Control Evaluation

Stress 𝐒𝐦 linearization Type Classification

Condition Intensity Value Result

(MPa) of stress symbol

SINT (MPa)

SI (Pm ) N/A 1.0×Sm =137 N/A

Design MEMBRANE

SII (PL ) 45.48 1.5×Sm =205.5 PASS

Conditions 137

A MEMBRANE PLUS

SIII (PL + Pb ) 59.05 1.5×Sm =205.5 PASS

BENDING

43

Figure 38: Stress variation through the thickness along Path-4

44

4.5.5 Path-5

Path-5 is the path along the thickness of the rectangular Nozzle around the middle section where

cylindrical shell body of the vessel connects the upper side of Nozzle N-1. Since this path is far

from the singularity, the average membrane stress and membrane plus bending stress along this

path are classified as primary and hence, checked against the 1.5×Sm limit.

Equivalent

Allowable ANSYS Stress

Load Stress Stress Control Evaluation

Stress 𝐒𝐦 linearization Type Classification

Condition Intensity Value Result

(MPa) of stress symbol

SINT (MPa)

SI (Pm ) N/A 1.0×Sm =137 N/A

Design MEMBRANE

Conditions 137 SII (PL ) 73.18 1.5×Sm =205.5 PASS

A MEMBRANE PLUS

SIII (PL + Pb ) 80.24 1.5×Sm =205.5 PASS

BENDING

45

Figure 40: Stress variation through the thickness along Path-5

46

Now we repeat the above stated paths for smaller rectangular Nozzle N-2

4.5.6 Path-6

Path-6 is the path along cylindrical shell thickness at the corner where the shell body of the vessel

connects rectangular nozzle-2. Since this path is near the singularity at the corner, the membrane

plus bending stress consists of primary as well as secondary stress components. Due to presence

of this secondary stress component, the allowable limit for total membrane plus bending stress

along this path is 3×Sm.

Equivalent

Allowable ANSYS Stress

Load Stress Stress Control Evaluation

Stress 𝐒𝐦 linearization Type Classification

Condition Intensity Value Result

(MPa) of stress symbol

SINT (MPa)

SI (Pm ) N/A 1.0×Sm =137 N/A

Design MEMBRANE

Conditions 137 SII (PL ) 61.36 1.5×Sm =205.5 PASS

A MEMBRANE PLUS

𝑆𝐼𝑉 (𝑃𝐿 + 𝑃𝑏 + 𝑄 ) 128.2 3×Sm =411 PASS

BENDING

47

Figure 42: Stress variation through the thickness along Path-6

48

4.5.7 Path-7

Path-7 is the path along Rectangular Nozzle thickness at the corner where shell body of the

pressure vessel connects smaller rectangular Nozzle N-2. Similar to the above described path-6,

this path also lies near the singularity and therefore the allowable limit for total membrane plus

bending stress along this path is 3×Sm.

Equivalent

Allowable ANSYS Stress

Load Stress Stress Control Evaluation

Stress 𝐒𝐦 linearization Type Classification

Condition Intensity Value Result

(MPa) of stress symbol

SINT (MPa)

SI (Pm ) N/A 1.0×Sm =137 N/A

Design MEMBRANE

Conditions 137 SII (PL ) 109.5 1.5×Sm =205.5 PASS

A MEMBRANE PLUS

𝑆𝐼𝑉 (𝑃𝐿 + 𝑃𝑏 + 𝑄 ) 169.1 3×Sm =411 PASS

BENDING

49

Figure 44: Stress variation through the thickness along Path-7

50

4.5.8 Path-8

Path-8 is the path along the thickness of the rectangular Nozzle at the middle section where

cylindrical shell body of the vessel connects the Nozzle N-2. This path lies far away from the

singularity. The average membrane stress and membrane plus bending stress along this path are

classified as primary and hence checked against the 1.5×Sm limit.

Equivalent

Allowable ANSYS Stress

Load Stress Stress Control Evaluation

Stress 𝐒𝐦 linearization Type Classification

Condition Intensity Value Result

(MPa) of stress symbol

SINT (MPa)

SI (Pm ) N/A 1.0×Sm =137 N/A

Design MEMBRANE

Conditions 137 SII (PL ) 16.72 1.5×Sm =205.5 PASS

A MEMBRANE PLUS

SIII (PL + Pb ) 26.88 1.5×Sm =205.5 PASS

BENDING

51

Figure 46: Stress variation through the thickness along Path-8

52

4.5.9 Path-9

Path-9 is the path along the thickness of the rectangular Nozzle around the middle section where

cylindrical shell body of the vessel connects the upper side of nozzle N-2. Since this path lies far

from the singularity, the average membrane stress and membrane plus bending stress along this

path are classified as primary and hence checked against the 1.5×Sm limit.

Equivalent

Allowable ANSYS Stress

Load Stress Stress Control Evaluation

Stress 𝐒𝐦 linearization Type Classification

Condition Intensity Value Result

(MPa) of stress symbol

SINT (MPa)

SI (Pm ) N/A 1.0×Sm =137 N/A

Design MEMBRANE

Conditions 137 SII (PL ) 44.89 1.5×Sm =205.5 PASS

A MEMBRANE PLUS

SIII (PL + Pb ) 54.67 1.5×Sm =205.5 PASS

BENDING

53

Figure 48: Stress variation through the thickness along Path-9

54

4.5.10 Path-10

Path-10 is the path through cylindrical shell of the vessel, in the section between the N1 nozzle

and N2 nozzle. This path is remote from the singularity region. The stresses along this path are

classified as primary and checked against the 1.5×Sm limit. This main reason for evaluating

stresses along this path is to check the strength of cylindrical shell body of the pressure vessel.

Equivalent

Allowable ANSYS Stress

Load Stress Stress Control Evaluation

Stress 𝐒𝐦 linearization Type Classification

Condition Intensity Value Result

(MPa) of stress symbol

SINT (MPa)

SI (Pm ) N/A 1.0×Sm =137 N/A

Design MEMBRANE

Conditions 137 SII (PL ) 7.97 1.5×Sm =205.5 PASS

A MEMBRANE PLUS

SIII (PL + Pb ) 57.81 1.5×Sm =205.5 PASS

BENDING

55

Figure 50: Stress variation through the thickness along Path-10

56

4.5.11 Path-11

Path-11 is the path created through the N-1 nozzle thickness at the joint between the flange and

N-1 nozzle. The membrane stress intensity linearized along this path is classified as primary local

stress and thereby will be verified against the ASME local stress limits which is equal

to 1.5×Sm(Allowable stress). For design to be safe, the linearized membrane stress along this

path should be less than 1.5×Sm(Allowable stress).

Equivalent

Allowable ANSYS Stress

Load Stress Stress Control Evaluation

Stress 𝐒𝐦 linearization Type Classification

Condition Intensity Value Result

(MPa) of stress symbol

SINT (MPa)

SI (Pm ) N/A 1.0×Sm =137 N/A

Design MEMBRANE

SII (PL ) 46.34 1.5×Sm =205.5 PASS

Conditions 137

A MEMBRANE PLUS

SIII (PL + Pb ) 77.29 1.5×Sm =205.5 PASS

BENDING

57

Figure 52: Stress variation through the thickness along Path-11

58

4.5.12 Path-12

Path-12 is a path created through the flange thickness at the joint between the flange and N-1

nozzle. This path is constructed on the flange at nozzle N-1 to check the safe design of the flange

component at the nozzle opening. This path is far from the singularity. The average membrane

stress and membrane plus bending stress along this path are classified as primary and checked

against the 1.5×Sm limit.

Equivalent

Allowable ANSYS Stress

Load Stress Stress Control Evaluation

Stress 𝐒𝐦 linearization Type Classification

Condition Intensity Value Result

(MPa) of stress symbol

SINT (MPa)

SI (Pm ) N/A 1.0×Sm =137 N/A

Design MEMBRANE

SII (PL ) 16.67 1.5×Sm =205.5 PASS

Conditions 137

A MEMBRANE PLUS

SIII (PL + Pb ) 47.77 1.5×Sm =205.5 PASS

BENDING

59

Figure 54: Stress variation through the thickness along Path-12

60

4.5.13 Path-13

Path-13 is a path through the nozzle thickness at the joint between the flange and N-2 small

nozzle. This path is created on nozzle N-2. Similar to path-11, this path is also far from the

singularity. Thus, the average membrane stress and membrane plus bending stress along this

path are classified as primary and hence checked against the 1.5×Sm limit.

Equivalent

Allowable ANSYS Stress

Load Stress Stress Control Evaluation

Stress 𝐒𝐦 linearization Type Classification

Condition Intensity Value Result

(MPa) of stress symbol

SINT (MPa)

SI (Pm ) N/A 1.0×Sm =137 N/A

Design MEMBRANE

Conditions 137 SII (PL ) 36.10 1.5×Sm =205.5 PASS

A MEMBRANE PLUS

SIII (PL + Pb ) 64.37 1.5×Sm =205.5 PASS

BENDING

61

Figure 56: Stress variation through the thickness along Path-13

62

4.5.14 Path-14

Path-14 is a path through the flange thickness at the joint between the flange and N-2 small

nozzle. This path is constructed on the flange connected to nozzle N-2. This path lies far from the

singularity. The average membrane stress and membrane plus bending stress along this path are

classified as primary and hence checked against the 1.5×Sm limit.

Equivalent

Allowable ANSYS Stress

Load Stress Stress Control Evaluation

Stress 𝐒𝐦 linearization Type Classification

Condition Intensity Value Result

(MPa) of stress symbol

SINT (MPa)

SI (Pm ) N/A 1.0×Sm =137 N/A

Design MEMBRANE

Conditions 137 SII (PL ) 10.39 1.5×Sm =205.5 PASS

A MEMBRANE PLUS

SIII (PL + Pb ) 30.87 1.5×Sm =205.5 PASS

BENDING

63

Figure 58: Stress variation through the thickness along Path-14

64

4.5.15 Paths-15,16,17,18,19 and 20

With stress linearization approach, it is preferred to have a stress classification line passing

through the stress singularity point in order to completely capture the effects of local stress

concentrations in this region. But, since our pressure vessel model contains filleted corner, we

cannot pin-point the exact location of the singularity. Thus, we cannot be assured whether the

paths path-2 and path-3 created above passes through the singularity. Hence, as a precautionary

measure and to ensure safe design of the pressure vessel, a few more paths were created around

the maximum stress region. Paths-15,17,19 were created through the thickness of the cylindrical

shell body of the vessel whereas Paths-16,18,20 were created through the thickness of Nozzle

N-1. The stress analysis results and verifications performed along these paths are shown below.

The linearized membrane plus bending stress along these paths are classified as secondary and

thereby will be verified against the ASME local stress limits which is equal to 3×Sm (Allowable

stress).

Figure 59: Additional paths created near the stress singularity region

65

Table 36: Path-15 Evaluation

Equivalent

Allowable ANSYS Stress

Load Stress Stress Control Evaluation

Stress 𝐒𝐦 linearization Type Classification

Condition Intensity Value Result

(MPa) of stress symbol

SINT (MPa)

SI (Pm ) N/A 1.0×Sm =137 N/A

Design MEMBRANE

Conditions 137 SII (PL ) 96.24 1.5×Sm =205.5 PASS

A MEMBRANE PLUS

𝑆𝐼𝑉 (𝑃𝐿 + 𝑃𝑏 + 𝑄 ) 198.1 3×Sm =411 PASS

BENDING

66

Table 37: Path-16 Evaluation

Equivalent

Allowable ANSYS Stress

Load Stress Stress Control Evaluation

Stress 𝐒𝐦 linearization Type Classification

Condition Intensity Value Result

(MPa) of stress symbol

SINT (MPa)

SI (Pm ) N/A 1.0×Sm =137 N/A

Design MEMBRANE

Conditions 137 SII (PL ) 158.1 1.5×Sm =205.5 PASS

A MEMBRANE PLUS

𝑆𝐼𝑉 (𝑃𝐿 + 𝑃𝑏 + 𝑄 ) 240.1 3×Sm =411 PASS

BENDING

67

Table 38: Path-17 Evaluation

Equivalent

Allowable ANSYS Stress

Load Stress Stress Control Evaluation

Stress 𝐒𝐦 linearization Type Classification

Condition Intensity Value Result

(MPa) of stress symbol

SINT (MPa)

SI (Pm ) N/A 1.0×Sm =137 N/A

Design MEMBRANE

Conditions 137 SII (PL ) 129.5 1.5×Sm =205.5 PASS

A MEMBRANE PLUS

𝑆𝐼𝑉 (𝑃𝐿 + 𝑃𝑏 + 𝑄 ) 227.7 3×Sm =411 PASS

BENDING

68

Table 39: Path-18 Evaluation

Equivalent

Allowable ANSYS Stress

Load Stress Stress Control Evaluation

Stress 𝐒𝐦 linearization Type Classification

Condition Intensity Value Result

(MPa) of stress symbol

SINT (MPa)

SI (Pm ) N/A 1.0×Sm =137 N/A

Design MEMBRANE

Conditions 137 SII (PL ) 151.1 1.5×Sm =205.5 PASS

A MEMBRANE PLUS

𝑆𝐼𝑉 (𝑃𝐿 + 𝑃𝑏 + 𝑄 ) 225.1 3×Sm =411 PASS

BENDING

69

Table 40: Path-19 Evaluation

Equivalent

Allowable ANSYS Stress

Load Stress Stress Control Evaluation

Stress 𝐒𝐦 linearization Type Classification

Condition Intensity Value Result

(MPa) of stress symbol

SINT (MPa)

SI (Pm ) N/A 1.0×Sm =137 N/A

Design MEMBRANE

Conditions 137 SII (PL ) 78.53 1.5×Sm =205.5 PASS

A MEMBRANE PLUS

𝑆𝐼𝑉 (𝑃𝐿 + 𝑃𝑏 + 𝑄 ) 174.3 3×Sm =411 PASS

BENDING

70

Table 41: Path-20 Evaluation

Equivalent

Allowable ANSYS Stress

Load Stress Stress Control Evaluation

Stress 𝐒𝐦 linearization Type Classification

Condition Intensity Value Result

(MPa) of stress symbol

SINT (MPa)

SI (Pm ) N/A 1.0×Sm =137 N/A

Design MEMBRANE

Conditions 137 SII (PL ) 159.4 1.5×Sm =205.5 PASS

A MEMBRANE PLUS

𝑆𝐼𝑉 (𝑃𝐿 + 𝑃𝑏 + 𝑄 ) 249.1 3×Sm =411 PASS

BENDING

71

4.6 Summary of Stress Analysis Results

Table 42: Stress Verification

Path

Type Category (MPa) (MPa)

Membrane PL 1.5×Sm 205.5 59.85

PATH-1

Membrane Plus Bending 𝑃𝐿 + 𝑃𝑏 1.5×Sm 205.5 73.26

Membrane PL 1.5×Sm 205.5 105.1

PATH-2

Membrane Plus Bending 𝑃𝐿 + 𝑃𝑏 + 𝑄 3×Sm 411 206.6

Membrane PL 1.5×Sm 205.5 157.1

PATH-3

Membrane Plus Bending 𝑃𝐿 + 𝑃𝑏 + 𝑄 3×Sm 411 235.0

Membrane PL 1.5×Sm 205.5 45.48

PATH-4

Membrane Plus Bending 𝑃𝐿 + 𝑃𝑏 1.5×Sm 205.5 59.05

Membrane PL 1.5×Sm 205.5 73.18

PATH-5

Membrane Plus Bending 𝑃𝐿 + 𝑃𝑏 1.5×Sm 205.5 80.24

Membrane PL 1.5×Sm 205.5 61.36

PATH-6

Membrane Plus Bending 𝑃𝐿 + 𝑃𝑏 + 𝑄 3×Sm 411 128.2

Membrane PL 1.5×Sm 205.5 109.5

PATH-7

Membrane Plus Bending 𝑃𝐿 + 𝑃𝑏 + 𝑄 3×Sm 411 169.1

Membrane PL 1.5×Sm 205.5 16.72

PATH-8

Membrane Plus Bending 𝑃𝐿 + 𝑃𝑏 1.5×Sm 205.5 26.88

Membrane PL 1.5×Sm 205.5 44.89

PATH-9

Membrane Plus Bending 𝑃𝐿 + 𝑃𝑏 1.5×Sm 205.5 54.67

Membrane PL 1.5×Sm 205.5 7.97

PATH-10

Membrane Plus Bending 𝑃𝐿 + 𝑃𝑏 1.5×Sm 205.5 57.81

Membrane PL 1.5×Sm 205.5 46.34

PATH-11

Membrane Plus Bending 𝑃𝐿 + 𝑃𝑏 1.5×Sm 205.5 77.29

Membrane PL 1.5×Sm 205.5 16.67

PATH-12

Membrane Plus Bending 𝑃𝐿 + 𝑃𝑏 1.5×Sm 205.5 47.77

Membrane PL 1.5×Sm 205.5 36.10

PATH-13

Membrane Plus Bending 𝑃𝐿 + 𝑃𝑏 1.5×Sm 205.5 64.37

Membrane PL 1.5×Sm 205.5 10.39

PATH-14

Membrane Plus Bending 𝑃𝐿 + 𝑃𝑏 1.5×Sm 205.5 30.87

Membrane PL 1.5×Sm 205.5 96.24

PATH-15

Membrane Plus Bending 𝑃𝐿 + 𝑃𝑏 + 𝑄 3×Sm 411 198.1

Membrane PL 1.5×Sm 205.5 158.1

PATH-16

Membrane Plus Bending 𝑃𝐿 + 𝑃𝑏 + 𝑄 3×Sm 411 240.1

Membrane PL 1.5×Sm 205.5 129.5

PATH-17

Membrane Plus Bending 𝑃𝐿 + 𝑃𝑏 + 𝑄 3×Sm 411 227.7

Membrane PL 1.5×Sm 205.5 151.1

PATH-18

Membrane Plus Bending 𝑃𝐿 + 𝑃𝑏 + 𝑄 3×Sm 411 225.1

Membrane PL 1.5×Sm 205.5 78.53

PATH-19

Membrane Plus Bending 𝑃𝐿 + 𝑃𝑏 + 𝑄 3×Sm 411 174.3

Membrane PL 1.5×Sm 205.5 159.4

PATH-20

Membrane Plus Bending 𝑃𝐿 + 𝑃𝑏 + 𝑄 3×Sm 411 249.1

72

From the stress analysis of the initial design, it is observed that the maximum membrane and

membrane plus bending stress occurs along path-20 whereas minimum membrane stress occurs

at path-10 and minimum membrane plus bending stress is found along path-8. We did expect the

maximum stress values along path-20, as this path/SCL was created in the high stress region of

the model. Note that, stress analysis was performed on nominal design of the pressure vessel

model with shell thickness of 10 mm and flange thickness of 16 mm. The above stress verification

results demonstrates that linearized stresses evaluated along all the paths are well-within the

allowable limits and hence this design of the pressure vessel model is legitimate and safe.

Although the initial design is valid, it is not the best fit solution as the weight of the pressure vessel

can be further reduced while it can still sustain stresses within the allowable limits. Thus, a

volume/weight minimizing optimization was carried out by varying the shell thickness and flange

thickness of the vessel. When we seek the optimal design for weight reduction of this pressure

vessel model, the above-mentioned stress linearization and classification approach is followed

and the stress verifications as specified by the ASME Pressure vessel code are evaluated for

every optimization search iteration. Note that, the linearization paths creation is also automated

through the APDL script file such that the paths are recreated in the same location of the geometry

every time the design parameters changes. This allows us to evaluate the stresses along the

same locations in the model for every search iteration in the optimization loop. The details of the

optimization problem formulation for the pressure vessel model in MATLAB are discussed in the

next chapter.

From the general standpoint of searching for the best available design, optimization can be

defined as follows. Mathematical optimization is the process of maximizing and/or minimizing one

or more objectives without violating specified design constraints, by regulating a set of variable

parameters that influence both the objectives and the design constraints. It is important to realize

that in order to apply mathematical optimization, you need to express the objective(s) and the

design constraint(s) as quantitative functions of the variable parameters. These variable

parameters are also known as design variables or decision variables.

MATLAB Optimization tool provides some very powerful functions for finding parameters that

minimize or maximize objectives while satisfying constraints. Based on the problem in hand, these

optimization solvers can be used along with a suitable algorithm to find the optimal solution. The

73

tool includes solvers for linear programming, mixed-integer linear programming, quadratic

programming, nonlinear optimization, and nonlinear least squares. We can use these solvers to

find optimal solutions to continuous and discrete problems, perform tradeoff analyses, and

incorporate optimization methods into algorithms and applications.

A general structural optimization problem can be mathematically formulated using the following

set of equations:

min f(x)

subject to

g(x) ≤ 0

h(x) = 0

where

xL ≤ x ≤ xU

The function f(x) represents the objective function or the cost function, which we would like to

minimize or maximize. The function g(x) represents a vector of inequality constraints evaluated

at x and the function h(x) represents a vector of equality constraints evaluated at x. The vector x

represents the vector of real-valued design variables. These are the quantities that we can change

in the design to improve its behavior. The constraints on the design variables, xL and xU , are

called side constraints. Design variables cannot be chosen arbitrarily; they must satisfy certain

specific functional requirements to produce an acceptable design. For example, in this pressure

vessel design, the variables selected by the optimization algorithm should be such that the design

passes the stress verifications in order to be considered as an acceptable design. These

restrictions that must be satisfied in a design are called design constraints. Design constraints are

classified into two; one that represent limitations on the behavior or performance of the system

and one that pose physical limitations on the design variables. While the former is referred to as

behavior or functional constraint, the latter is known as geometric or side constraints. An efficient

and accurate solution to the above stated optimization problem depends not only on the size of

the problem in terms of the number of constraints and design variables but also on characteristics

of the objective function and constraints.

For the pressure vessel model under consideration, the above shown optimization problem can

be mathematically formulated as:

74

minimize: Total Weight of the Pressure Vessel f(x)

subject to

Primary Membrane Stress 𝑃𝐿 ≤ 205.5 MPa

Primary Membrane Plus Bending Stress 𝑃𝐿 + 𝑃𝑏 ≤ 205.5 MPa g(x)

Primary Membrane Plus Bending + Secondary Stress 𝑃𝐿 + 𝑃𝑏 + 𝑄 ≤ 411 MPa

h(x) = 0

where

4 mm ≤ x(1): shell thickness ≤ 10 mm

16 mm ≤ x(2): flange thickness ≤ 30 mm

Objective Function: The objective function to be minimized is the total weight of the pressure

vessel. Like most of the conventional optimization problems, this is also a single objective

optimization problem. Instead of the conventional method where the objective in MATLAB is

evaluated as a function of design parameters, a scalar value representing the total volume of the

pressure vessel structure, obtained from ANSYS FEA is being directly feed into the objective

function value.

considered to manufacture it efficiently by meeting up the industry requirements. For the analysis

of the current pressure vessel equipment, we consider the shell thickness and flange thickness

of the vessel as the design parameters. During Optimization, both the design parameters are

varied in a specified range. The shell thickness value is varied from 4mm to 10mm, whereas the

flange thickness value is varied from 16mm to 30mm. The figure-66 displayed below, shows the

design variables used in this optimization problem.

75

Constraints: In general, almost all engineering problems are constrained. Constraints can be

either inequality constraints (g(x) ≤ or g(x) ≥) or equality constraints (h(x) = 0). The feasible

region for inequality constraints represents the entire area on the feasible side of the allowable

value; for equality constraints, the feasible region is only where the constraint is equal to the

allowable value. As discussed in section 4.2, different allowable limits have been specified for

membrane and membrane plus bending stress, based on the stress category and path locations.

For the current pressure vessel optimization, these stress limits form the inequality design

constraints. Note that, this optimization problem does not contain equality constraints. As

mentioned earlier, the design should comply with these constraints in order to be considered as

a feasible solution.

Side Constraints: Side constraints can be described as the geometrical or physical limitations

imposed on the design variables. Since this is a minimization problem, the range of the design

variables are the side constraints. In MATLAB, the side constraints are defined by two sets of

vectors xL and xU where xL represents a lower bound on the design variables and xU represents

the upper bound on the design variables. Number of elements on each of these vectors must be

equal to the number of design variables. A lower bound of 4 mm has been set for the shell

thickness because of the geometrical restrictions. When shell thickness has a value lower than

4mm, some of the geometrical entities in the finite element model becomes invalid and thus the

analysis can no longer be conducted. An upper bound of 10 mm was set for the shell thickness.

Similarly, an a lower bound of 16mm and an upper bound of 30 mm was set for flange thickness.

• Choose an optimization solver.

• Create an objective function, typically the function you want to minimize.

• Create constraints, if any.

• Set options, or use the default options.

• Call the appropriate solver/Algorithm.

Once we have defined the optimization problem, we are ready to start searching the design space.

Prior to optimizing a design, it is useful to employ design space exploration—a quantitative

method that help engineers gain a better, more complete understanding of a structure's potential

by discovering which design variables will have the greatest impact on the structure's

76

performance. Design exploration assumes that the optimal design is initially unknown and initially

uncharacterizable. The process of design exploration discovers design conditions and through

experimentation characterizes what an optimal design looks like. Once this is known, the final

solution can then be found through a convergent design optimization algorithm. The essential

quantitative method for design space exploration is design-of-experiment (DOE) studies. In a

DOE study, an analysis model is automatically evaluated multiple times, with the design variables

set to different values in each iteration. The results identify which variable(s) affect the design the

most, and which least.

A design of experiments study was conducted for the finite element model of the pressure vessel.

This analysis model is automatically evaluated multiple times, with the design variables – shell

thickness and flange thickness set to different values in each iteration. The results of DOE process

were then used to generate the response of the model. Analyzing response of a system is

necessary for visualizing the design space, examining relationships among design variables and

their effects on key responses, and rapidly evaluating design alternatives. The figure-67 below

depicts the response of the pressure vessel model generated in MATLAB. The plot is based on

data created by meshing the space with a 20X20 grid of values for the design parameters, shell

and flange thickness. Hence the finite element model of the pressure vessel was evaluated 400

times to generate the results for the contour plot. The shell thickness parameter was varied from

4mm to 10mm, whereas flange thickness parameter was varied from 16mm to 30mm.

It is important to keep in mind that the graphical representation is typically restricted to two

variables. For three variables, we need a fourth dimension to resolve the information while three-

dimensional contour plots are not easy to illustrate. Sometimes, even the three-dimensional

graphical representation does not really enhance our understanding of the problem or solution.

But since we are considering only two design variables for the current optimization problem of

pressure vessel, we can easily plot and analyze the graphical results.

77

Figure 67: Response of the Pressure Vessel Model

It can be seen from figure-67 that total weight of the pressure vessel reduces linearly with the

decrease in shell and flange thickness. Smooth flat continuous surface in the response indicates

a linearly varying objective function, weight of the vessel in this case.

representation of the design functions involved in the problem. In order to determine the feasible

design region for our parameters, we need to draw the contours of the objective function and

design constraints. Since, as per the DOE results the maximum membrane and membrane plus

bending stress occurs along path-17, we will use stress results along this path to draw the

constraint contours. The figure-68 below represents the design space created in MATLAB with

the same data points which were used to create the response. The membrane and membrane

plus bending stress contours are shown on the same plot, drawn in blue and red color

respectively. The solid black lines represent contours of the total volume of the pressure vessel,

the objective. These contour plots give the insight on the behavior of the objective function with

respect to change in the design variables. We can see that weight decreases as we move from

78

the upper right corner of the design space towards the lower left corner. According to the design

space, optimal solution should lie in the range of shell thickness 4mm-4.5mm and flange thickness

of 22mm-30mm. This is a feasible region for the optimum solution because the pressure vessel

model can sustain the stresses within the allowable stress limits for membrane and membrane

plus bending stress, when design parameter values lie in this specific region. If we look closely at

figure-68, the possible space for optimal solution is drawn on the design space.

79

Figure 69: Membrane Stress Contours Figure 70: Membrane Plus Bending Stress Contours

Figure-69 and figure-70 represents the stress contours for the linearized membrane and

membrane plus bending stress evaluated along path-17. The contours shows the increase in

stress with decrease in shell thickness and flange thickness values.

For the nominal value of shell thickness equal to 10 millimeters and flange thickness equal to 16

millimeters, the total volume possessed by the pressure vessel was evaluated to be 198559610

cubic millimeters and the maximum element Von-Mises stress was calculated to be 325.77 MPa.

As this is a minimization problem, our aim should be to achieve a lower value for the total weight

(objective function) in comparison with the nominal value. Again, remember that the linearized

stresses along all the paths must be less than the allowable stress limits to avoid failure.

Individual graphs were plotted to determine the level of influence that each design variable poses

on the objective function. The figure-71 shows the variation of the total volume of the vessel with

respect to the change in shell thickness while the flange thickness was fixed to the nominal value

of 16mm. The figure-72 shows the variation of the total volume of the vessel with respect to the

change in flange thickness while the shell thickness was fixed to the nominal value of 10mm. It

can be clearly seen from the figures that our design function, i.e., total weight of the pressure

vessel is more sensitive to change in shell thickness parameter. This was in-fact expected

because shell thickness parameter has more weightage in the model geometry construction in

comparison to flange thickness.

80

Figure 71: Objective function (Total Volume) vs shell Figure 72: Objective function (Total Volume) vs flange

thickness thickness

The MATLAB provides inbuilt functions like fminbnd, fminunc, fminsearch, ga, linprog, quadprog,

fmincon, lsqcurvefit, fgoalattain, lsqnonlin, etc which can be used for minimization, multiobjective

optimization, and solving least-squares or data-fitting problems. The most difficult part in MATLAB,

is to decide which solver and algorithm will produce the best optimal results for the problem in

hand. For the current problem, ‘fmincon’ is the best suitable option because this is a minimization

problem which involves a constrained nonlinear multivariable function. fmincon is a MatLab inbuilt

nonlinear solver for optimization which best applies to smooth objective functions with smooth

constraints. It is a gradient-based method that is designed to work on problems where the

objective and constraint functions are both continuous and have continuous first derivatives. It

has proved to be a suitable tool to solve many optimization problems in the mechanical

engineering field. Like most of the optimization solvers, fmincon only guarantee's finding a local

optimum. To run fmincon, we need to define the objective function, constraint function and we

also need to specify the initial design variable values. The user can also set the optimization

options for fmincon or any other solver in MATLAB by using the ‘optimoptions’ function.

The fmincon function in MATLAB is called using the command shown below:

[Xopt,Fopt] = fmincon(fun,x0,A,b,Aeq,beq,lb,ub,nonlcon,options)

81

where x0, A, b, Aeq, beq, LB, and UB are the input variables that need to be defined before calling

fmincon. ‘fun’ is the name of the function file containing the definition of f(x), and ‘nonlcon’ is the

name of the function file containing the nonlinear constraints. The variables Xopt and Fopt are

the outputs of fmincon, where Xopt is the optimum vector of variables [x1,x2] and Fopt is the

minimum value of the objective function. Other output details like exit flag, stopping criteria

message and gradient values can also be extracted. A, b, Aeq, and beq: These variables need to

be defined only if the problem has linear constraints. In many cases, all constraints (linear and

nonlinear) can be defined in the nonlcon.m file, so these variables can simply be defined as empty

matrices. LB and UB are the vectors that define lower and upper bounds on the design variables.

fmicon starts the optimization at x0 and attempts to find a minimizer x of the function described in

fun subjected to the linear and nonlinear, equality and in-equality constraints.

The optimset command can be used to set or change the values of the optimization options. Some

of these options are relevant to particular algorithms. The options arguments include algorithms

selection, stopping criteria, iteration display settings, step tolerance, constraint tolerance, Max

iterations, Max function evaluations, plot functions etc. The function optimset creates an options

structure that is passed as an input argument to the optimization solver (fmincon in this case). For

the pressure vessel optimization problem, a few options that were changed from their default

value are explained below in table-43:

Default

Option Name Description Changed Value

Value

Algorithm used by solver

Algorithm Interior-point Active-set

(fmincon)

Minimum change in variables

DiffMinChange 1e-6 0.1

for finite differencing

Termination tolerance on x,

the current point. TolX is a

tolx lower bound on the size of a

1e-6 1e-4

(Step Tolerance) step. If the solver attempts to

take a step that is smaller

than TolX, the iterations end.

Maximum number of

MaxIter 100 10000

iterations allowed.

Maximum number of

MaxFunEvals 200 10000

function evaluations allowed.

82

{@optimplotx,

User-defined or built-in plot

@optimplotfval,

function that an optimization @optimplotconstrviolation,

PlotFcns {}

function calls at each @optimplotfunccount,

@optimplotfirstorderopt,

iteration. @optimplotstepsize}

Level of display. 'off' displays

no output; 'iter' displays

output at each iteration;

Display 'final' displays just the final off iter

output; 'notify' displays

output only if the function

does not converge.

After defining the above quantities, the function fmincon is called. The function fmincon calls (i)

nonlcon.m to evaluate the constraints and (ii) fun.m to evaluate the objective function. fmincon

provides five different algorithms options:

1. 'interior-point' (default)

2. 'Trust-region-reflective'

3. 'sqp'

4. 'sqp-legacy'

5. 'active-set'

In the present work, active-set and sqp algorithms are used. Results are discussed at the end.

The other algorithms are excluded because either they are time consuming or they are not

suitable for the problem in hand. For example, implementing Trust-Region-Reflective Algorithm

is very complex process as it requires user specified gradient for both objective and constraint

functions. Understanding how these algorithms work requires advanced statistics and machine

learning background and since this is beyond the scope of this work, only a brief description of

these algorithms is stated below.

Interior- Point: 'interior-point' handles large, sparse problems, as well as small dense problems.

The algorithm satisfies bounds at all iterations, and can recover from NaN or Inf results. It is a

large-scale algorithm; The algorithm can use special techniques for large-scale problems. For

details, see Interior-Point Algorithm in fmincon options

83

Active-set: 'active-set' can take large steps, which adds speed. The algorithm is effective on

some problems with non-smooth constraints. It is not a large-scale algorithm. Lagrange multipliers

are directly computed based on the solution of KKT (Karush-Kuhn-Tucker) equations.

Constrained quasi-Newton methods guarantee superlinear convergence by accumulating

second-order information regarding the KKT equations using a quasi-Newton updating procedure.

Like sqp algorithm, a QP sub-problem is solved at each major iteration.

Sqp: 'sqp' satisfies bounds at all iterations. It is not a large-scale algorithm. This method allows

us to closely mimic Newton's method for constrained optimization just as is done for

unconstrained optimization. At each major iteration, an approximation is made of the Hessian of

the Lagrangian function using a quasi-Newton updating method. This is then used to generate a

QP subproblem whose solution is used to form a search direction for a line search procedure.

This algorithm has proved to be superior in terms of efficiency, accuracy, and percentage of

successful solutions, over a large number of test problems.

only bounds or linear equality constraints, but not both. Within these limitations, the algorithm

handles both large sparse problems and small dense problems efficiently. It is a large-scale

algorithm. The algorithm can use special techniques to save memory usage, such as a Hessian

multiply function.

As discussed in previous section, the objective function, i.e., total weight of the pressure vessel

is more sensitive to change in shell thickness. Hence, to effectively reduce the weight of the

vessel, optimization algorithms would try minimize the shell thickness of the vessel as much as

possible. On the other hand, since the maximum stress occurs along the corner junction of shell

body and nozzle, the algorithms would try maximize the flange thickness in order to comply with

the stress constraints.

In this chapter, the methodology to interface ANSYS and MATLAB is explained in detail. It is worth

highlighting that any other finite element software either licensed or opened source may be used

to be coupled with MatLab. The requirement to be fulfilled is that the software must allow

programming the finite element model by means of a script file in order to be able to automate

the proposed optimization methodology shown. This kind of integrated approach is desired in

optimization of complex designs because it is completely automated and does not require any

84

kind of user intervention, until an optimum solution is found. The coupling between MatLab and

the finite element software is done by means of a batch command shown below.

-i C:\Users\artik\Desktop\Thesis\PressureVesselModel_script.txt

-o C:\Users\artik\Desktop\Thesis\FEAreport.txt

Where -b is the batch command followed by -i and -o which represent the directory for input and

output file respectively. The Objective function file “PressureVessel_obj.m” defined in MatLab

calls ANSYS to runs in batch mode. This file updates the variables in the Ansys script file

“PressureVesselModel_script.txt”, executes ANSYS in batch mode and evaluates the Objective

function and stress results.

The main advantage of creating and solving a model by means of APDL script is that the model

can be defined in terms of variables, thus creating a parametric model. The variables that are

employed to create a parametric model for the pressure vessel are shell thickness and flange

thickness. The values of these variables will be varied by the optimization algorithm until a

minimum is reached. APDL post-process allows, to store the results of the analyzed model in a

text file. Required data are extracted from this text file and are fed to the optimization tool in order

to redefine the design variables.

Figure-73 below defines the optimization loop. It can be seen that the information between

ANSYS and MatLab is exchanged based on the text files which are overwritten in each loop. It

should be noted that all the created files must be placed on the same directory of the hard drive

or else the user should set appropriate path for these files.

The file “PressureVessel_Optimization.m” is the main file which runs fmincon to evaluate

the objective function. (Weight of the pressure Vessel in this case).

As soon as the main file is run in MATLAB, the objective function file

‘PressureVessel_obj.m’ is called.

Objective function file runs ANSYS in batch mode by using the above-mentioned

command.

ANSYS script file ‘PressureVesselModel_script.txt’ retrieves the values of shell thickness

and flange thickness from MATLAB design variables and the finite element model is

solved in ANSYS.

85

Results obtained from finite element analysis are stored in a text file “ANYSresults.txt”.

These results are extracted in MATLAB to evaluate the Objective function and constraints.

fmincon algorithm updates the value of design parameters and the loop is repeated until

an optimum solution is found.

7.1 Results

Several cases were run by calling fmincon with different initial design variable values and with

different step tolerance limits, to check the convergence of the optimal solution. The optimized

results obtained by using active-set algorithm were compared with the results given by sqp

algorithm. The first three runs were conducted using active-set algorithm whereas all the other

runs were performed using sqp algorithm. As shown in table-45 below, the best solution was given

86

by run-2 in which the total weight of the pressure vessel was reduced by about 39.21%. The

convergence of the optimal solution was checked using different initial design variable values and

different step tolerance values. From table-44, we can see that a faster convergence is obtained

by using lower step tolerance values with a slight decrease in the weight reduction %. The reason

behind this is that for tight tolerances, fmincon evaluates more number of points which thereby

increases the number of iterations taken to get convergence at the optimal point. Hence the best

approach is to start optimization with a low step tolerance value whenever the problem is time

consuming and the design space is widely spread. It should be noted that since fmincon is not a

global optimization solver, optimized values of the design variables are possible local minimum.

The optimization results along with the descriptive iterations for different fmincon runs are shown

below:

Fmincon run-1

X0(2) =flange thickness (Unit: mm)

Lower and Upper Bounds on Design parameters: XL=[4 16] XU=[10 30]

'MaxIter', 10000, 'MaxFunEvals', 10000, 'TolX' ,1e-2, 'PlotFcns',

{@optimplotx,@optimplotfval,@optimplotconstrviolation,@optimplotfunccount,@optimplotfirstor

deropt,@optimplotstepsize}, 'display', 'iter')

198559610 − 122236730

Total Volume Reduction = ×100 = 38.43%

198559610

87

Figure 74: fmincon run-1 iterations

88

Fmincon run-2

X0(2) =flange thickness (Unit: mm)

Lower and Upper Bounds on Design parameters: XL=[4 16] XU=[10 30]

'MaxIter', 10000, 'MaxFunEvals', 10000, 'TolX' ,1e-3, 'PlotFcns',

{@optimplotx,@optimplotfval,@optimplotconstrviolation,@optimplotfunccount,@optimplotfirstor

deropt,@optimplotstepsize}, 'display', 'iter')

198559610 − 120699140

Total Volume Reduction = ×100 = 39.21%

198559610

89

Figure 77: Plot Functions for fmincon run-2

Fmincon run-3

Initial Design Points: X0= [5 23] where,

X0(1) =shell thickness

X0(2) =flange thickness

Lower and Upper Bounds on Design parameters: XL=[4 16] XU=[10 30]

'MaxIter', 10000, 'MaxFunEvals', 10000, 'TolX' ,1e-4, 'PlotFcns',

{@optimplotx,@optimplotfval,@optimplotconstrviolation,@optimplotfunccount,@optimplotfirstor

deropt,@optimplotstepsize}, 'display', 'iter')

198559610 − 121669870

Total Volume Reduction = ×100 = 38.72%

198559610

90

Figure 78: fmincon run-3 iterations

91

Figure 79: Plot Functions for fmincon run-3

Initial Design Points: X0= [4.1 23] where,

X0(1) =shell thickness

X0(2) =flange thickness

Lower and Upper Bounds on Design parameters: XL=[4 16] XU=[10 30]

Optimization Settings: optimoptions(@fmincon, 'Algorithm', ‘sqp’, DiffMinChange’, 0.01,

'MaxIter', 10000, 'MaxFunEvals', 10000, 'TolX' ,1e-2, 'PlotFcns',

{@optimplotx,@optimplotfval,@optimplotconstrviolation,@optimplotfunccount,@optimplotfirstor

deropt,@optimplotstepsize}, 'display', 'iter')

198559610 − 123065150

Total Volume Reduction = ×100 = 38.02%

198559610

92

Figure 80: fmincon run-4 iterations

93

Fmincon run-5 (sqp-algorithm)

Initial Design Points: X0= [4.8 29] where,

X0(1) =shell thickness

X0(2) =flange thickness

Lower and Upper Bounds on Design parameters: XL=[4 16] XU=[10 30]

Optimization Settings: optimoptions(@fmincon, 'Algorithm', ‘sqp’, DiffMinChange’, 0.01,

'MaxIter', 10000, 'MaxFunEvals', 10000, 'TolX' ,1e-2, 'PlotFcns',

{@optimplotx,@optimplotfval,@optimplotconstrviolation,@optimplotfunccount,@optimplotfirstor

deropt,@optimplotstepsize}, 'display', 'iter')

198559610 − 128726400

Total Volume Reduction = ×100 = 35.17%

198559610

Initial Design Points: X0= [4.8 29] where,

X0(1) =shell thickness

X0(2) =flange thickness

Lower and Upper Bounds on Design parameters: XL=[4 16] XU=[10 30]

Optimization Settings: optimoptions(@fmincon, 'Algorithm', ‘sqp’, DiffMinChange’, 0.01,

'MaxIter', 10000, 'MaxFunEvals', 10000, 'TolX' ,1e-3, 'PlotFcns',

94

{@optimplotx,@optimplotfval,@optimplotconstrviolation,@optimplotfunccount,@optimplotfirstor

deropt,@optimplotstepsize}, 'display', 'iter')

198559610 − 125423580

Total Volume Reduction = ×100 = 36.83%

198559610

95

Figure 84: Plot functions for fmincon run-6

Initial Design Points: X0= [4.8 29] where,

X0(1) =shell thickness

X0(2) =flange thickness

Lower and Upper Bounds on Design parameters: XL=[4 16] XU=[10 30]

Optimization Settings: optimoptions(@fmincon, 'Algorithm', ‘sqp’, DiffMinChange’, 0.01,

'MaxIter', 10000, 'MaxFunEvals', 10000, 'TolX' ,1e-4, 'PlotFcns',

{@optimplotx,@optimplotfval,@optimplotconstrviolation,@optimplotfunccount,@optimplotfirstor

deropt,@optimplotstepsize}, 'display', 'iter')

198559610 − 125356370

Total Volume Reduction = ×100 = 36.86%

198559610

96

Figure 85: fmincon run-7 iterations

97

Figure 86: Plot Functions for fmincon run-7

Fmincon run-8

Initial Design Points: X0= [5 23] where,

X0(1) =shell thickness

X0(2) =flange thickness

Lower and Upper Bounds on Design parameters: XL=[4 16] XU=[10 30]

Optimization Settings: optimoptions(@fmincon, 'Algorithm', ‘active-set’, DiffMinChange’, 0.1,

'MaxIter', 10000, 'MaxFunEvals', 10000, 'TolX' ,1e-3, 'PlotFcns',

{@optimplotx,@optimplotfval,@optimplotconstrviolation,@optimplotfunccount,@optimplotfirstor

deropt,@optimplotstepsize}, 'display', 'iter')

198559610 − 121959420

Total Volume Reduction = ×100 = 38.58%

198559610

98

Figure 87: fmincon run-8 iterations

99

Table 44: fmincon Optimization Convergence for different step tolerance limits

Run-1 Run-8 Run-3 Run-5 Run-6 Run-7

Step tolerance ‘tolx’ 1e-2 1e-3 1e-4 1e-2 1e-3 1e-4

Initial Design

Variable X0

[5 23] [5 23] [5 23] [4.1 29] [4.1 29] [4.1 29]

Optimized Design [4.1609 [4.1442 [4.1338 [4.0000 [4.0000 [4.0000

Variable Xopt 23.0116] 22.9901] 22.8975] 28.9607] 26.7502] 26.7051]

Number of iterations

taken

7 14 66 2 17 23

Optimized Design

Design Constraints

Variables Optimal

Weight

Objective: Maximum Maximum

fmincon x1: shell x2: flange Reduction

Total Volume Membrane Membrane

thickness thickness %

(mm3) stress Plus Bending

(mm) (mm)

(MPa) stress (MPa)

Initial Design 10 16 198559610 NA 159.4 249.1

Run-1 4.1609 23.0116 122236730 38.43 205.30 340.67

Active- Run-2 4.0000 23.5703 120699140 39.21 204.70 337.73

set Run-3 4.1338 22.8975 121669870 38.72 205.46 338.07

Run-8 4.1442 22.9901 121959420 38.58 205.35 350.07

Run-4 4.2339 22.8461 122918140 38.02 205.05 353.59

Run-5 4.0000 28.9607 128726400 35.17 191.19 296.37

sqp

Run-6 4.0000 26.7502 125423580 36.83 204.29 313.46

Run-7 4.0000 26.7051 125356370 36.86 204.02 304.88

Best

Run-2 4.0000 23.5703 120699140 39.21 204.70 337.73

Solution

7.2 Conclusions

The total weight of the pressure vessel was reduced by 39%. The optimal design parameters for

the pressure vessel are obtained and the objective minimization of cost by reducing weight of the

Pressure vessel is achieved. Design parameters, shell thickness and flange thickness, are

optimized while limiting the maximum linearized membrane and membrane plus bending stresses

100

below the allowable limits. Based on the comparison of different fmincon runs, weight reduction

obtained by using active-set algorithm was in the range of 38% - 39% whereas the reduction

obtained by using sqp algorithm was in the range of 35% - 37%. As predicted from the design

space analysis, the optimal point in fact does lies in the estimated optimal solution region. Hence,

it can be concluded that this kind of interfacing approach where powerful FEA softwares such as

ANSYS are integrated with robust optimization tools like MATLAB, can be used to solve different

types of complex design optimization problems. It was also found that the optimization in design

of pressure vessel using FEA is a safe and promising method as it has successfully satisfied the

goal of weight reduction. The methodology implemented in this study is very effective and can be

a successful tool for advance analysis in the structural design field. The integrated approach used

in this thesis work is desired for solving optimization problems because the process is completely

automated and does not require any kind of user intervention until an optimum solution is found.

Optimization solution for finite element model of this pressure vessel is mesh dependent, but

since we have a discontinuity in the model, the mesh density was kept fixed throughout the

optimization process. In particular, the optimization problem was solved by using MATLAB inbuilt

optimization solver called 'fmincon'. As FMINCON gives the local minimum within the limits

specified for the design variables, the minimum weight value obtained in this study is a local

minimum.

This pressure vessel structure was also modeled in ANSYS workbench with the purpose

of performing the optimization using workbench’s goal-driven optimization tool. But due to

time constraint, this study could not be conducted.

The analysis can be reexamined by modeling the thin cylindrical shell body of the pressure

vessel using shell elements. Optimized results obtained from this shell model investigation

can be compared and verified with the results obtained from the current work.

processing including meshing the finite element model in Hypermesh, solving the model

in ANSYS, followed by the optimization in MATLAB. Altair Hypermesh provides superior

meshing options with good element level control compared to all FEA softwares

101

available in the market. This will definitely give a much accurate solution since user

can exploit the desired unique features limited to each software.

Further analysis can be conducted on this pressure vessel by considering a few more

design parameter such as radius of the cylindrical shell body, length of the cylindrical

shell body, head thickness, head height, height of the nozzle openings etc.

Metaheuristic based global optimization algorithms like genetic algorithm, Ant colony

optimization algorithm, Differential Evolution and Simulated Annealing can be used to find

the global optimum for the pressure vessel model used in this work.

from specific components of the pressure vessel.

REFERENCES

1. A. Gauchía, B.L. Boada, M.J.L. Boada and V. Díaz (2014). Integration of MATLAB and

ANSYS for Advanced Analysis of Vehicle Structures, MATLAB Applications for the

Practical Engineer, Mr. Kelly Bennett (Ed.), InTech, DOI: 10.5772/57390. Available

from:http://www.intechopen.com/books/matlab-applications-for-the-practical-

engineer/integration-of-matlab-and-ansys-for-advanced-analysis-of-vehicle-structures

NOZZLE TO PRESSURE VESSEL CONNECTION FINITE ELEMENT MODELS,

published in Pressure Vessel and Piping Codes and Standards - 2000 PVPVol . 407,

ed. A. F. Deardorff, p. 271-275, ISBN No. 0-7918-1888-8.

3. Carlos A. de J. Miranda, Altair A. Faloppa, Miguel Mattar Neto and Gerson Fainer,

“ASME STRESS LINEARIZATION AND CLASSIFICATION – A DISCUSSION BASED

ON A CASE STUDY”, 2011 International Nuclear Atlantic Conference - INAC 2011Belo

Horizonte, MG, Brazil, October 24-28, 2011 ISBN: 978-85-99141-04-5.

of pressure vessels using shape optimization: An integrated approach”, International

Journal of Pressure Vessels and Piping, Volume 88, May 2011.

5. Prof. Vishal V. Saidpatil and Prof. Arun S. Thakare, "Design & Weight Optimization of

Pressure Vessel Due to Thickness Using Finite Element Analysis”, International Journal

of Emerging Engineering Research and Technology Volume 2, Issue 3, June 2014, PP

1-8 ISSN 2349-4395 (Print) & ISSN 2349-4409 (Online).

102

6. Sulaiman Hassan, Kavi Kumar, Ch Deva Raj and Kota Sridhar, "Design and

Optimisation of Pressure Vessel Using Metaheuristic Approach", Applied Mechanics

and Materials Vols. 465-466 (2014) pp401-406, Trans Tech Publications, Switzerland.

CYLINDRICAL PRESSURE VESSEL USING FEA", International Journal of Scientific

and Research Publications, Volume 5, Issue 12, December 2015, ISSN 2250-3153.

Boilers", ASME Boiler and Pressure Vessel Comitte July-1 2010 New York.

9. Arturs Kalnins, "Stress Classification in Pressure Vessels and piping", Pressure Vessel

and Piping Systems.

10. "The Nuts and Bolts of Stress Linearization", Pressure Vessel Engineering Ltd.

11. Achille Messac, "Optimization in Practice with MATLAB for Engineering Students and

Professionals", published by CAMBRIDGE University press NY-2015.

12. Alan R. Parkinson, Richard J. Balling and John D. Hedengren, "Optimization Methods

for Engineering Design", 2013 Brigham Young University.

REGULATED MULTIQUADRIC RESPONSE SURFACE MODEL", Thesis Report, The

University of Texas at Arlington, December 2005.

14. Miguel Mattar Neto, Carlos Alexandre de Jesus Miranda, Altair Antonio Faloppa

and Gerson Fainer, "ASME Stress Linearization and Classification of a WYE

piping juction”, 2011 ANSYS Conference & ESSS Users Meeting.

APPENDIX B - GUIDELINES FOR ANALYSIS, IN-VESSEL COMPONENTS

https://www.mathworks.com/help/optim/ug/constrained-nonlinear-optimization-

algorithms.html

95.htm

103

19. Stress Categories for Design by Analysis of Pressure Vessels, Talk for ANSYS

USERS GROUP - June 1999

https://www.sharcnet.ca/Software/Ansys/17.0/enus/help/wb_sim/ds_linearized_stress

es.html

III & VIII, Interpretation of Finite Element Analysis Stress Results, Rev 4, 2000

APPENDIX

HOPPER DIAGRAGM

104

BIBLIOGRAPHY

Artik Patel enrolled in PES Institute of Technology, Bangalore, India in 2008 and earned his B.E.

(Bachelor of Engineering) in Mechanical Engineering in June 2012. After his graduation, he

worked as a Project Engineer at Wipro technologies, Bangalore, India from August 2012 to April

2014. He had started his M.S. (Master of Science) degree in Mechanical Engineering at The

University of Texas at Arlington, Arlington, Texas in August 2014 and earned M.S. Degree in

Mechanical Engineering in December 2016 with a 4.0 G.P.A. During his tenure as a master’s

student, he served as the Graduate Teaching Assistant of Dr. Raul Fernandez, Dr. Adrian

Rodriguez and Dr. Kent Lawrence. His thesis was based on the design optimization of pressure

vessel structure subjected to design constraints specified by the ASME Boiler and Pressure

Vessel code. His current research interests are finite elements, structural optimization and

computer aided design. His long-term goal is to work as a structural analyst in the engineering

industry.

105

- Full Text 01Transféré parsakthivel
- Fem notepadTransféré parVivek Khare
- A15. Method for Multi-objective Optimized Designs of Surface Mount Permanent Magnet Motors with Concentrated or Distributed Stator.pdfTransféré parwalidghoneim1970
- 02_2011_077_Croccollo_04Transféré parKirti Patil
- Fem Application BiomechanicsTransféré parAbhigyan Keshri
- INTRODUCTION TO FINITE ELEMENT METHOD.docTransféré parashishphadke1972
- FEA_DEMOTransféré pardharminderarora
- 10.1.1.832.5384Transféré parCarmen Servigna
- FEA SummaryTransféré parsharma_ankur_jec
- 10.1.1.163.1989Transféré paronderali
- b17 Handout 4Transféré parKristinochka Slavenkovna
- ijnaoe-2013-0226Transféré parArturo
- General Procedure of FEA 7Transféré partauasi_1
- Me Less Sssssssss SssTransféré paretayhailu
- Will Poorly-Shaped Elements Really Affect My Solution_ _ CAE AssociatesTransféré pargaliskribd
- 639-1862-1-PBTransféré pardhanashekarV
- Ansys Lab ManualTransféré parideepuj
- IFEM.apporigins of FEMTransféré parsunny045
- 4I16-IJAET0916854_v6_iss4_1464to1473.pdfTransféré parkaviarasu1990
- Matlab-toolbookTransféré parLeonel Gzz
- 1999a Finite Element Method for CrackTransféré parJaime Jose Batista
- CpedTransféré parJeanne Roselle Dulatre Cortez
- IFEM.Ch06Transféré parjeroldscd
- Staticki izvestajTransféré parTonic89
- form8.pdfTransféré parnathamar100
- Finite Element AnalysisTransféré parTesisTraduccionesRuzel
- Finite ElementTransféré parClaudia Garnica Trujillo
- Coupled Boundary Element Finite Element AnalysiTransféré parSaif Rahman
- Finite Element Analysis of Aircraft Wing Using Composite StructureTransféré parAumair Malik
- Soil Structure Interaction Calculus, For Rigid Hydraulic Structures, Using FEMTransféré parIOSRjournal

- 2002DO_FelicioBruzziBarros 2002Transféré parbruno18
- NMIR80-1Transféré parManasses junior
- materials-12-00609.pdfTransféré parManasses junior
- 723610b94d7db590143bca1f3a2c1710.pdfTransféré parManasses junior
- 723610b94d7db590143bca1f3a2c1710.pdfTransféré parManasses junior
- LIVRO_U2 (2)Transféré parManasses junior
- Compliance Report as 28091Transféré parManasses junior
- resistenciacorrosaoTransféré parjjsearom
- Good)Transféré parshrinidhee
- Kat_Linie-VW_en.pdfTransféré parManasses junior
- catalogo Gates correia.pdfTransféré parManasses junior
- Trabalho de vasos_2009_1Transféré parrobinho_301083rob
- Motorcycle chain engineeringTransféré parSagar Dhuri
- 1812-1812-1-PBTransféré parManasses junior
- ASEE Allen Paper Revised Version Mar 12Transféré parManasses junior
- DriveLineWhitePape.pdfTransféré parManasses junior
- Curriculum Revision MarchTransféré parManasses junior
- Structural Analysis by FEM using ANSYSTransféré parSteransko
- Marc_2017.1_training_120_Workbook_636522147316424295.pdfTransféré parManasses junior
- 6P-QM36G-Engenharia Economica e FinancasTransféré parManasses junior
- Cantilevered Beam Using 1d and 2d ElementsTransféré parManasses junior
- SERIAL NUMBER.txtTransféré parAhmed Ezwai
- Anchor AnalysisTransféré parManasses junior
- Gruas e Mini Gruas Problemas Montagem e OperaçãoTransféré parAnonymous fGUP5h2A
- Connecting Rod Using 2d ElementsTransféré parManasses junior
- Revista Pontes de Aco Set2015Transféré parFred
- DSADSDSaaassW dSA .txtTransféré parManasses junior
- Geometric Dimensioning and Tolerancing Gd t Integration Throughout a Manufacturing Engineering CurriculumTransféré parhiren_mistry55
- 101497 Manual Lsf Engenharia 2016Transféré parpaulo
- Relatorio de Atividades 2018Transféré parManasses junior

- United States v. Gary Maden, 134 F.3d 383, 10th Cir. (1998)Transféré parScribd Government Docs
- 288310_633958949115308750Transféré paretyjrsdsd
- Kia ManualTransféré parRoger Sego
- BDO Seidman v. Hirshberg - NY CoA - 1999Transféré paremptywindow
- Integrated Marketing Communication (IMC) Unilever (Rexona)Transféré parBohMay Yip
- Torredes vs. Villamor _digested.docxTransféré parbelle_chary28
- Congos AsTransféré parmartha suclupe
- Linearizing RTDTransféré parcheng_chan
- Start 26Transféré parkPrasad8
- Documentary Film MovementTransféré parGeraldine Huaylinos Atuncar
- Textile Industry InformationTransféré parJasvinder Singh IT Trainer
- FACCGN in Las Vegas, Nevada - New FPACC Chapter Chamber Conducts Major SummitTransféré parLorna Dietz
- 12517-9425Transféré parcomphome13
- Khashayar Bagheri MBA, ADMS, BEng, IEng, EngTech, MIET (DBA Doctoral Candidate)Transféré parinfo9372
- Knight Frank Technical Due Diligence March 2013Transféré parChris Gonzales
- Report about Delta Modulation --Abdullah Azyem Ali 20156191.pdfTransféré parabdullah azayem
- (1976) Logan's Run (Press Kit)_textTransféré parrtaylor4792
- Kosta Stojanovic - Economic Problems of SerbiaTransféré parIgor Zdravkovic
- Gas Aggregation Company of Nigeria Investor Forum PresentationTransféré parsegunoyes
- Cew Moda Atterbury (2)Transféré parranger9013
- 440v, 50hz, 75-100ka Sirius Load Feeder en-usTransféré parerkamlakar2234
- @Parmak Freze TablolarTransféré parSerhat Tüysüz
- Caso Exito Governo Povo de Acre EnTransféré parPep Diz
- Chapter 2 - Discrete Time Signals and SystemsTransféré pardvishwakarma1796
- CISSP_2004Transféré parSakil Mahmud
- epa 3Transféré parjanfarsa9623
- Overall Equipment EffectivenessTransféré parMarciano Zanotto
- self-guided social media training manual (1)Transféré parapi-272460715
- Annual Report 2014-15Transféré parRahul Roy
- Rank ChangerTransféré parpoopidoo